OXYGENHUB
◆ v0.4.3 oxygen / crates / oxy-core / src / repo / mod.rs
97.6 KiB · Rust · last changed Push a long history in parts 26 days ago source blame history raw
98286406 Andy McCall ↶ 1 
//! A repository on disk: the `.oxy` directory, one workspace, the operation
2 
//! log, and the working-copy change (spec §12, §13.1).
3 
//!
83e0677b Andy McCall ↶ 4 
//! Not yet implemented from the spec: packs, checkpoints, multiple
5 
//! workspaces, multi-signer policy updates, and the `objects`/`checkpoint`
6 
//! endpoints. A repository without a policy is unverified and accepts
7 
//! unsigned pushes; with one, everything received is checked (spec §14.5). Changes are signed individually when `core.sign` is
98286406 Andy McCall ↶ 8 
//! `change`; the default `seal` leaves them unsigned until push exists.
9
10 
use crate::attest::{Attest, AttestBody};
11 
use crate::diff::{ranges, split_lines, text_edits};
12 
use crate::hash::{hash_object, ChangeHash, KeyId, ObjectRef, ObjectType};
7ba155f2 Andy McCall ↶ 13 
use crate::identity::{
14 
    verify as verify_signature, Delegate, Domain, Identity, Scope, SecretKey, Signature,
15 
};
98286406 Andy McCall ↶ 16 
use crate::ids::{ChangeId, NodeId, WorkspaceId};
75167672 Andy McCall 17 
use crate::objects::{
18 
    Blob, Change, ChangeKind, Edit, Hlc, Mode, NodeKind, OpRef, StateObject, TreeOp,
19 
};
98286406 Andy McCall ↶ 20 
use crate::oplog::{Format1Op, Op, View, SNAPSHOT_SUFFIX};
75167672 Andy McCall 21 
use crate::protocol::{
22 
    open_remote, Bundle, RefResult, RefUpdate, RefsResponse, RemoteOptions, RemoteRef,
23 
    SignedAttest, SignedPolicy,
24 
};
98286406 Andy McCall ↶ 25 
use crate::state::{FileContent, Snapshot, State};
26 
use crate::store::{Store, StoreError};
75167672 Andy McCall 27 
#[allow(unused_imports)]
28 
use crate::subrepo::Pin;
98286406 Andy McCall ↶ 29 
use crate::tree::OpKey;
30 
use rand::RngCore;
cc7a0056 Andy McCall ↶ 31 
use std::collections::{BTreeMap, BTreeSet, HashMap, HashSet};
98286406 Andy McCall ↶ 32 
use std::fs;
33 
use std::io::Write;
34 
use std::path::{Path, PathBuf};
35
36 
pub const OXY_DIR: &str = ".oxy";
37 
pub const FORMAT_LINE: &str = "oxy repository format 2\n";
38 
const FORMAT1_LINE: &str = "oxy repository format 1\n";
39 
const BINARY_SNIFF: usize = 8 * 1024;
d02e545d Andy McCall ↶ 40 
/// Loose objects tolerated before a commit packs them (spec §13.4).
41 
pub const AUTOPACK_THRESHOLD: usize = 4096;
98286406 Andy McCall ↶ 42
43 
#[derive(Debug, thiserror::Error)]
44 
pub enum RepoError {
45 
    #[error("io: {0}")]
46 
    Io(#[from] std::io::Error),
47 
    #[error(transparent)]
48 
    Store(#[from] StoreError),
49 
    #[error(transparent)]
50 
    Object(#[from] crate::objects::ObjectError),
51 
    #[error(transparent)]
52 
    State(#[from] crate::state::StateError),
53 
    #[error("{0}")]
54 
    Msg(String),
e266b7d4 Andy McCall ↶ 55 
    /// The operation would leave a conflict where none may be (a rebased
56 
    /// change cannot hold one); the caller may offer another way.
57 
    #[error("{0}")]
58 
    Conflicted(String),
98286406 Andy McCall ↶ 59 
}
60
61 
pub type Result<T> = std::result::Result<T, RepoError>;
62
dd21fd82 Andy McCall ↶ 63 
/// The operation head's reference record on a store that is not this
64 
/// disk's directory (oxygen#68).
65 
pub const HEAD_RECORD: &str = "ops/head";
66
67 
/// How often recording an operation reads the head again when other
68 
/// processes keep moving it.
69 
const HEAD_TRIES: usize = 8;
70
d069701a Andy McCall ↶ 71 
/// The object store of the repository directory `oxy`: `.oxy/objects`,
72 
/// or the store `storage.url` names behind a cache of `storage.cache`
73 
/// bytes in `.oxy/cache` (oxygen#68). `.oxy/objects` is then only
74 
/// scratch space: packs being written, and derived caches.
75 
fn open_objects(oxy: &Path, config: &Config) -> Result<Store> {
76 
    let Some(url) = &config.storage_url else {
77 
        return Ok(Store::open(oxy.join("objects"))?);
78 
    };
79 
    let options = crate::protocol::RemoteOptions {
80 
        cache_dir: Some(oxy.join("cache")),
81 
        ..Default::default()
82 
    };
83 
    let remote = crate::protocol::open_store(url, &options)
84 
        .ok_or_else(|| {
85 
            RepoError::Msg(format!(
86 
                "storage.url {url}: this oxy cannot open that kind of store (dir://, or s3://, gs:// and az:// in builds with cloud stores)"
87 
            ))
88 
        })?
89 
        .map_err(|e| RepoError::Msg(format!("open the repository's store {url}: {e}")))?;
90 
    let cached = crate::strata::CachingStore::open(
91 
        std::sync::Arc::from(remote),
92 
        oxy.join("cache"),
93 
        config.storage_cache,
94 
    )
95 
    .map_err(|e| RepoError::Msg(format!("open the cache for {url}: {e}")))?;
96 
    Ok(Store::open_on(
97 
        std::sync::Arc::new(cached),
98 
        oxy.join("objects"),
99 
    )?)
100 
}
101
98286406 Andy McCall ↶ 102 
fn fail<T>(s: impl Into<String>) -> Result<T> {
103 
    Err(RepoError::Msg(s.into()))
104 
}
105
7b9809e6 Andy McCall ↶ 106 
mod bookmarks;
98286406 Andy McCall ↶ 107 
mod checkout;
252e8cb2 Andy McCall ↶ 108 
mod checkpoint;
98286406 Andy McCall ↶ 109 
mod config;
3e5772cc Andy McCall ↶ 110 
mod doctor;
387335ff Andy McCall ↶ 111 
mod fsck;
ba160f60 Andy McCall ↶ 112 
mod gc;
79ab8528 Andy McCall ↶ 113 
mod git;
614541c3 Andy McCall ↶ 114 
mod history;
6d3eefbc Andy McCall ↶ 115 
mod hooks;
84b9e9ac Andy McCall ↶ 116 
mod hosted;
75167672 Andy McCall 117 
mod identity;
957fee30 Andy McCall ↶ 118 
mod inspect;
98a1620a Andy McCall ↶ 119 
mod ops;
83e0677b Andy McCall ↶ 120 
mod policy;
135212de Andy McCall ↶ 121 
mod prompt;
fc1a466a Andy McCall ↶ 122 
mod receipts;
47ce8b8e Andy McCall ↶ 123 
mod revset;
98286406 Andy McCall ↶ 124 
mod snapshot;
9f9e72b9 Andy McCall ↶ 125 
pub mod strata;
75167672 Andy McCall 126 
mod subrepos;
98286406 Andy McCall ↶ 127 
mod sync;
132b884b Andy McCall ↶ 128 
mod tags;
98286406 Andy McCall ↶ 129 
#[cfg(test)]
130 
mod tests;
170bb436 Andy McCall ↶ 131 
mod touched;
76740263 Andy McCall ↶ 132 
mod walk;
75167672 Andy McCall 133 
mod workspace;
98286406 Andy McCall ↶ 134
14edbaaf Andy McCall ↶ 135 
pub use bookmarks::{remote_bookmark, BookmarkInfo, MergeReport, RemoteOnlyBookmark, StateGap};
98286406 Andy McCall ↶ 136 
pub use checkout::CheckoutReport;
b3c5a59a Andy McCall ↶ 137 
pub use config::{parse_size, read_config_file, Config};
3e5772cc Andy McCall ↶ 138 
pub use doctor::{Check, DoctorReport, Health};
387335ff Andy McCall ↶ 139 
pub use fsck::{FsckReport, Problem, Severity};
ba160f60 Andy McCall ↶ 140 
pub use gc::GcReport;
62794ecc Andy McCall ↶ 141 
pub use git::{GitExportReport, GitImportReport, GitVerifyReport};
75167672 Andy McCall 142 
pub use hooks::{json_array, json_str, HookOutcome};
84b9e9ac Andy McCall ↶ 143 
pub use hosted::FileEdit;
28fc23d7 Andy McCall ↶ 144 
pub use identity::{Rotation, SignatureStatus};
6a81dbab Andy McCall ↶ 145 
pub use inspect::{BisectStep, BlameLine, Description};
98a1620a Andy McCall ↶ 146 
pub use ops::{RestoreReport, RevertReport, ViewChange};
28fc23d7 Andy McCall ↶ 147 
pub use policy::PolicyUpdate;
75167672 Andy McCall 148 
pub use prompt::{PromptInfo, DEFAULT_PROMPT_FORMAT, GLYPH_PROMPT_FORMAT};
fc1a466a Andy McCall ↶ 149 
pub use receipts::ReceiptInfo;
47ce8b8e Andy McCall ↶ 150 
pub use revset::{Revset, REVSET_BUDGET};
75167672 Andy McCall 151 
pub use snapshot::{
152 
    added_node_paths, diff_files_of, has_conflict_markers, is_binary, referenced_changes,
153 
    render_change, render_hunks, similarity, ChangeStatus, DiffHunk, FileDiff, Hunk, PathStatus,
154 
    Selection, SnapshotResult,
155 
};
156 
pub use subrepos::{SubrepoInfo, SubrepoSync};
cb8b8082 Andy McCall ↶ 157 
pub use sync::{FetchReport, PullReport, PushReport, RefUpdated, OBJECTS_RESULT};
132b884b Andy McCall ↶ 158 
pub use tags::TagInfo;
6caacfff Andy McCall ↶ 159 
pub use workspace::{FileLock, SnapshotLock, WorkingCopyLock, WorkspaceInfo};
98286406 Andy McCall ↶ 160
54cd08a1 Andy McCall ↶ 161 
/// The file name under which per-bookmark or per-tag data is kept: `/`
162 
/// (and `%`) escaped, so `feature/x` is one file beside `feature` rather
163 
/// than a directory that would clash with it.
164 
pub(crate) fn name_file(name: &str) -> String {
165 
    name.replace('%', "%25").replace('/', "%2F")
166 
}
167
168 
/// The name a [`name_file`] file name stands for.
169 
pub(crate) fn file_name_to_name(file: &str) -> String {
170 
    file.replace("%2F", "/").replace("%25", "%")
171 
}
172
98286406 Andy McCall ↶ 173 
fn write_atomic(path: &Path, data: &[u8]) -> Result<()> {
4dcb8c09 Andy McCall ↶ 174 
    Ok(crate::durable::write_atomic(path, data)?)
98286406 Andy McCall ↶ 175 
}
176
f9466021 Andy McCall ↶ 177 
/// Replace a cache file: whole or not at all, but without waiting for the
178 
/// disk. A cache checks its own bytes when read, so one a crash leaves
179 
/// short or stale is rebuilt rather than trusted.
180 
/// How many changes a state index may fall behind its working copy before
181 
/// it is written again; until then each load replays them.
182 
const INDEX_REPLAY_LIMIT: usize = 16;
acadb5ed Andy McCall ↶ 183 
/// The changes whose encodings are kept for building graphs: a batch of
184 
/// files often takes inline text from a few changes in turn.
185 
const CHANGE_CACHE_COUNT: usize = 4;
186 
const CHANGE_CACHE_BYTES: usize = 512 << 20;
638fb214 Andy McCall ↶ 187 
/// How many state indexes a workspace keeps.
188 
const STATE_INDEXES_KEPT: usize = 2;
f9466021 Andy McCall ↶ 189 
const INDEX_REPLAY_TIME: std::time::Duration = std::time::Duration::from_millis(50);
190
191 
pub(crate) fn write_cache(path: &Path, data: &[u8]) -> Result<()> {
192 
    let tmp = path.with_extension(format!("tmp-{}", std::process::id()));
193 
    fs::write(&tmp, data)?;
194 
    fs::rename(&tmp, path)?;
195 
    Ok(())
196 
}
197
98286406 Andy McCall ↶ 198 
pub struct LogEntry {
199 
    pub hash: ChangeHash,
2863456a Andy McCall ↶ 200 
    /// Metadata from the state; the full change may be absent in a shallow repository.
201 
    pub change: crate::state::ChangeMeta,
98286406 Andy McCall ↶ 202 
    pub is_working_copy: bool,
203 
    pub is_head: bool,
204 
}
205
992467d3 Andy McCall ↶ 206 
/// A text file's base graph key, by path, as a walk asks for it.
1966f5db Andy McCall ↶ 207 
pub(crate) type KeyOf<'a> = dyn Fn(&str) -> Option<[u8; 32]> + Sync + 'a;
992467d3 Andy McCall ↶ 208
c3069537 Andy McCall ↶ 209 
/// A file as found on disk or in an imported tree.
23ff1288 Andy McCall ↶ 210 
#[derive(Clone)]
c3069537 Andy McCall ↶ 211 
pub(crate) struct FsEntry {
212 
    pub kind: NodeKind,
213 
    pub exec: bool,
b7f102fc Andy McCall ↶ 214 
    /// File content, or a symlink's target. Empty when `blob` is set.
40a40bb7 Andy McCall ↶ 215 
    /// Shared, so a new file's insert holds it without a copy.
216 
    pub content: std::sync::Arc<Vec<u8>>,
b7f102fc Andy McCall ↶ 217 
    /// The stored blob of an opaque file whose size and mtime are unchanged
218 
    /// since the last manifest (spec §12.3 stat cache); the content was not
219 
    /// read.
220 
    pub blob: Option<ObjectRef>,
221 
    /// BLAKE3 of the content, carried over from the manifest when the stat
222 
    /// is unchanged, so the manifest can be rewritten without rehashing.
1966f5db Andy McCall ↶ 223 
    pub content_hash: Option<[u8; 32]>,
cc7a0056 Andy McCall ↶ 224 
    /// A text file known equal to the base's rendering without reading it:
225 
    /// its stat matches the manifest and the manifest recorded it equal to
226 
    /// a base file graph with the same key (spec §12.3). `content` is empty.
227 
    pub unchanged: bool,
228 
    /// The base graph key this file equals, for the next manifest.
1966f5db Andy McCall ↶ 229 
    pub graph_key: Option<[u8; 32]>,
0554aaa7 Andy McCall ↶ 230 
    /// Size and mtime already known to be current (a checkout leaving the
231 
    /// file as it was), so the manifest need not stat it again.
232 
    pub stat: Option<(u64, u128)>,
98286406 Andy McCall ↶ 233 
}
234
b7f102fc Andy McCall ↶ 235 
impl FsEntry {
236 
    pub(crate) fn new(kind: NodeKind, exec: bool, content: Vec<u8>) -> FsEntry {
75167672 Andy McCall 237 
        FsEntry {
238 
            kind,
239 
            exec,
40a40bb7 Andy McCall ↶ 240 
            content: std::sync::Arc::new(content),
75167672 Andy McCall 241 
            blob: None,
242 
            content_hash: None,
cc7a0056 Andy McCall ↶ 243 
            unchanged: false,
244 
            graph_key: None,
0554aaa7 Andy McCall ↶ 245 
            stat: None,
75167672 Andy McCall 246 
        }
b7f102fc Andy McCall ↶ 247 
    }
248 
}
249
98286406 Andy McCall ↶ 250 
pub struct Repo {
4dee277c Andy McCall ↶ 251 
    /// Oxygen-pinned subrepos skipped by the running Git export.
252 
    pub(crate) git_export_skipped: std::cell::Cell<usize>,
d0990c2b Andy McCall ↶ 253 
    /// The manifest as last read, with the file identity it was read from,
254 
    /// so one command parses it once (see `read_manifest`).
255 
    pub(crate) manifest_cache: std::cell::RefCell<Option<prompt::CachedManifest>>,
76740263 Andy McCall ↶ 256 
    /// A listing of the working copy left by a failed clean check, for the
257 
    /// snapshot that follows it to reuse; taken by the next walk and
258 
    /// dropped whenever the manifest is written.
259 
    pub(crate) listing_cache: std::cell::RefCell<Option<walk::Listing>>,
d4dbab97 Andy McCall ↶ 260 
    /// The nested repositories the last walk left out, being neither
261 
    /// registered nor tracked subrepos (oxygen#43).
262 
    pub(crate) untracked_nested: std::cell::RefCell<Vec<String>>,
98286406 Andy McCall ↶ 263 
    root: PathBuf,
264 
    oxy: PathBuf,
265 
    store: Store,
266 
    workspace: WorkspaceId,
e02bf0cd Andy McCall ↶ 267 
    /// A hosted repository with no working copy (spec §12.2, hub spec §4.1).
268 
    bare: bool,
98286406 Andy McCall ↶ 269 
    keys_dir: PathBuf,
270 
    user_config: PathBuf,
271 
    pub config: Config,
c863c7a7 Andy McCall ↶ 272 
    /// What the working copy's file system can represent (spec §7.7).
273 
    platform: crate::portable::Platform,
acadb5ed Andy McCall ↶ 274 
    /// The encodings of the changes last read for their inline inserts,
275 
    /// oldest first.
276 
    last_change: std::cell::RefCell<Vec<(ChangeHash, std::sync::Arc<Vec<u8>>)>>,
98a1620a Andy McCall ↶ 277 
    /// `--at-op`: read the repository as it was after this operation;
278 
    /// recording an operation is refused.
279 
    at_op: Option<ObjectRef>,
1ec3f008 Andy McCall ↶ 280 
    /// The directories (with their times) and nested repositories of the
281 
    /// last walk, for the manifest written from it.
282 
    pub(crate) walked_dirs: std::cell::RefCell<Option<WalkedDirs>>,
0389c3e8 Andy McCall ↶ 283 
    /// The directories a manifest check listed, for its restamp only.
284 
    pub(crate) listed_dirs: std::cell::RefCell<Option<WalkedDirs>>,
98286406 Andy McCall ↶ 285 
}
286
1ec3f008 Andy McCall ↶ 287 
/// A walk's directories with their times, and its nested repositories.
288 
pub(crate) type WalkedDirs = (Vec<(String, u128)>, Vec<String>);
289
e02bf0cd Andy McCall ↶ 290 
/// Marker file of a bare repository, inside `.oxy/`.
291 
const BARE_MARKER: &str = "bare";
75167672 Andy McCall 292 
/// Marker of a hosted repository that has no id yet: the first push binds
293 
/// it to the pusher's id (spec §16.1).
294 
const UNBOUND_MARKER: &str = "unbound";
e02bf0cd Andy McCall ↶ 295
98286406 Andy McCall ↶ 296 
/// The user's oxy directory: `$OXY_HOME`, else `$XDG_CONFIG_HOME/oxy`, else
297 
/// `~/.config/oxy`. It holds `config.toml` (spec §19) and `keys/`.
298 
pub fn default_home() -> PathBuf {
299 
    if let Ok(h) = std::env::var("OXY_HOME") {
300 
        return PathBuf::from(h);
301 
    }
302 
    if let Ok(x) = std::env::var("XDG_CONFIG_HOME") {
303 
        return PathBuf::from(x).join("oxy");
304 
    }
abaa8d51 Andy McCall ↶ 305 
    // Windows has no HOME as a rule: the roaming profile's application
306 
    // data, as Windows programs keep theirs.
307 
    #[cfg(windows)]
308 
    if let Some(d) = std::env::var_os("APPDATA").or_else(|| std::env::var_os("USERPROFILE")) {
309 
        return PathBuf::from(d).join("oxy");
310 
    }
98286406 Andy McCall ↶ 311 
    let home = std::env::var("HOME").unwrap_or_else(|_| ".".into());
312 
    PathBuf::from(home).join(".config").join("oxy")
313 
}
314
315 
pub fn default_keys_dir() -> PathBuf {
316 
    default_home().join("keys")
317 
}
318
319 
pub fn default_user_config() -> PathBuf {
320 
    default_home().join("config.toml")
321 
}
322
323 
/// Where `create_identity` records the author.
324 
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
325 
pub enum IdentityScope {
326 
    /// The user file: every repository of this user signs with it.
327 
    User,
328 
    /// This repository only.
329 
    Repo,
330 
}
331
332 
impl Repo {
333 
    // ------------------------------------------------------------ lifecycle
334
335 
    pub fn init(root: &Path) -> Result<Repo> {
336 
        let mut id = [0u8; 16];
337 
        rand::rngs::OsRng.fill_bytes(&mut id);
338 
        Self::init_with_id(root, id)
339 
    }
340
341 
    /// Initialize a repository with a given repository ID (clones share the
342 
    /// ID of their origin, spec §14.4).
343 
    pub fn init_with_id(root: &Path, id: [u8; 16]) -> Result<Repo> {
d069701a Andy McCall ↶ 344 
        Self::init_on(root, id, None)
345 
    }
346
347 
    /// A new repository with a fresh id, bare or not, on `storage` or in
348 
    /// `.oxy/objects` (`oxy init [--bare] [--store URL]`).
349 
    pub fn init_new(root: &Path, bare: bool, storage: Option<&str>) -> Result<Repo> {
350 
        let mut id = [0u8; 16];
351 
        rand::rngs::OsRng.fill_bytes(&mut id);
352 
        if bare {
353 
            Self::init_bare_on(root, id, storage)
354 
        } else {
355 
            Self::init_on(root, id, storage)
356 
        }
357 
    }
358
359 
    /// Initialize a repository whose objects and operation head are kept
360 
    /// on the store at `storage` (`storage.url`), or in `.oxy/objects`.
361 
    pub fn init_on(root: &Path, id: [u8; 16], storage: Option<&str>) -> Result<Repo> {
98286406 Andy McCall ↶ 362 
        fs::create_dir_all(root)?;
363 
        let root = root.canonicalize().unwrap_or_else(|_| root.to_path_buf());
364 
        let oxy = root.join(OXY_DIR);
365 
        if oxy.exists() {
366 
            return fail(format!("{} is already an oxy repository", root.display()));
367 
        }
368 
        fs::create_dir_all(oxy.join("ops"))?;
369 
        write_atomic(&oxy.join("format"), FORMAT_LINE.as_bytes())?;
370 
        write_atomic(&oxy.join("id"), hex::encode(id).as_bytes())?;
d069701a Andy McCall ↶ 371 
        let mut config = Config::load(Some(&default_user_config()), &oxy.join("config"))?;
372 
        if let Some(url) = storage {
373 
            config.set("storage.url", url)?;
374 
        }
98286406 Andy McCall ↶ 375 
        config.save(&oxy.join("config"))?;
376 
        let workspace = WorkspaceId::random();
377 
        write_atomic(&oxy.join("workspace"), workspace.to_hex().as_bytes())?;
d069701a Andy McCall ↶ 378 
        let store = open_objects(&oxy, &config)?;
75167672 Andy McCall 379 
        let repo = Repo {
519c7c06 Andy McCall ↶ 380 
            last_change: Default::default(),
98a1620a Andy McCall ↶ 381 
            at_op: None,
1ec3f008 Andy McCall ↶ 382 
            walked_dirs: Default::default(),
0389c3e8 Andy McCall ↶ 383 
            listed_dirs: Default::default(),
75167672 Andy McCall 384 
            git_export_skipped: Default::default(),
d0990c2b Andy McCall ↶ 385 
            manifest_cache: Default::default(),
76740263 Andy McCall ↶ 386 
            listing_cache: Default::default(),
d4dbab97 Andy McCall ↶ 387 
            untracked_nested: Default::default(),
75167672 Andy McCall 388 
            root,
389 
            oxy,
390 
            store,
391 
            workspace,
392 
            bare: false,
393 
            keys_dir: default_keys_dir(),
c863c7a7 Andy McCall ↶ 394 
            platform: crate::portable::Platform::current(),
75167672 Andy McCall 395 
            user_config: default_user_config(),
396 
            config,
397 
        };
4e88e774 Andy McCall ↶ 398 
        repo.register_workspace(&workspace, &repo.root)?;
98286406 Andy McCall ↶ 399
400 
        let wc = repo.new_working_copy(vec![], "");
401 
        let wc_hash = repo.store.put_change(&wc)?;
402 
        let mut view = View::default();
75167672 Andy McCall 403 
        view.bookmarks.insert(
404 
            repo.config.bookmark.clone(),
405 
            repo.store.put_state(&StateObject::new(vec![]))?,
406 
        );
98286406 Andy McCall ↶ 407 
        view.workspaces.insert(workspace, wc_hash);
408 
        repo.record_op(&[], "oxy init", view)?;
409 
        Ok(repo)
410 
    }
411
e02bf0cd Andy McCall ↶ 412 
    /// Create a bare repository: an object store, operation log and
413 
    /// references with no workspace and no working copy, for hosting
414 
    /// (hub spec §4.1). `root` holds `.oxy/` and nothing else.
415 
    pub fn init_bare(root: &Path) -> Result<Repo> {
416 
        let mut id = [0u8; 16];
417 
        rand::rngs::OsRng.fill_bytes(&mut id);
418 
        Self::init_bare_with_id(root, id)
419 
    }
420
421 
    pub fn init_bare_with_id(root: &Path, id: [u8; 16]) -> Result<Repo> {
d069701a Andy McCall ↶ 422 
        Self::init_bare_on(root, id, None)
423 
    }
424
425 
    /// [`Repo::init_on`] for a bare repository.
426 
    pub fn init_bare_on(root: &Path, id: [u8; 16], storage: Option<&str>) -> Result<Repo> {
e02bf0cd Andy McCall ↶ 427 
        fs::create_dir_all(root)?;
428 
        let root = root.canonicalize().unwrap_or_else(|_| root.to_path_buf());
429 
        let oxy = root.join(OXY_DIR);
430 
        if oxy.exists() {
431 
            return fail(format!("{} is already an oxy repository", root.display()));
432 
        }
433 
        fs::create_dir_all(oxy.join("ops"))?;
434 
        write_atomic(&oxy.join("format"), FORMAT_LINE.as_bytes())?;
435 
        write_atomic(&oxy.join("id"), hex::encode(id).as_bytes())?;
436 
        write_atomic(&oxy.join(BARE_MARKER), b"")?;
d069701a Andy McCall ↶ 437 
        let mut config = Config::load(Some(&default_user_config()), &oxy.join("config"))?;
438 
        if let Some(url) = storage {
439 
            config.set("storage.url", url)?;
440 
        }
e02bf0cd Andy McCall ↶ 441 
        config.save(&oxy.join("config"))?;
d069701a Andy McCall ↶ 442 
        let store = open_objects(&oxy, &config)?;
75167672 Andy McCall 443 
        let repo = Repo {
519c7c06 Andy McCall ↶ 444 
            last_change: Default::default(),
98a1620a Andy McCall ↶ 445 
            at_op: None,
1ec3f008 Andy McCall ↶ 446 
            walked_dirs: Default::default(),
0389c3e8 Andy McCall ↶ 447 
            listed_dirs: Default::default(),
75167672 Andy McCall 448 
            git_export_skipped: Default::default(),
d0990c2b Andy McCall ↶ 449 
            manifest_cache: Default::default(),
76740263 Andy McCall ↶ 450 
            listing_cache: Default::default(),
d4dbab97 Andy McCall ↶ 451 
            untracked_nested: Default::default(),
75167672 Andy McCall 452 
            root,
453 
            oxy,
454 
            store,
455 
            workspace: WorkspaceId::ZERO,
456 
            bare: true,
457 
            keys_dir: default_keys_dir(),
c863c7a7 Andy McCall ↶ 458 
            platform: crate::portable::Platform::current(),
75167672 Andy McCall 459 
            user_config: default_user_config(),
460 
            config,
461 
        };
e02bf0cd Andy McCall ↶ 462 
        let mut view = View::default();
75167672 Andy McCall 463 
        view.bookmarks.insert(
464 
            repo.config.bookmark.clone(),
465 
            repo.store.put_state(&StateObject::new(vec![]))?,
466 
        );
e02bf0cd Andy McCall ↶ 467 
        repo.record_op(&[], "oxy init --bare", view)?;
468 
        Ok(repo)
469 
    }
470
75167672 Andy McCall 471 
    /// A bare repository with no id yet. Its `id()` is all zero until the
472 
    /// first push binds it (`bind_id`), so an existing repository can be
473 
    /// pushed to a freshly created hosted one and keep its identity.
474 
    pub fn init_bare_unbound(root: &Path) -> Result<Repo> {
475 
        let repo = Self::init_bare_with_id(root, [0; 16])?;
476 
        fs::remove_file(repo.oxy.join("id"))?;
477 
        write_atomic(&repo.oxy.join(UNBOUND_MARKER), b"")?;
478 
        Ok(repo)
479 
    }
480
481 
    pub fn is_unbound(&self) -> bool {
482 
        self.oxy.join(UNBOUND_MARKER).is_file()
483 
    }
484
485 
    /// Give an unbound repository its id.
486 
    pub fn bind_id(&self, id: [u8; 16]) -> Result<()> {
487 
        if !self.is_unbound() {
488 
            return fail("this repository already has an id");
489 
        }
490 
        if id == [0; 16] {
491 
            return fail("a repository id cannot be all zero");
492 
        }
493 
        write_atomic(&self.oxy.join("id"), hex::encode(id).as_bytes())?;
494 
        fs::remove_file(self.oxy.join(UNBOUND_MARKER))?;
495 
        Ok(())
496 
    }
497
e02bf0cd Andy McCall ↶ 498 
    /// Whether this repository has no working copy (`init_bare`).
499 
    pub fn is_bare(&self) -> bool {
500 
        self.bare
501 
    }
502
98286406 Andy McCall ↶ 503 
    /// Open the repository containing `start` (searching upwards).
504 
    pub fn open(start: &Path) -> Result<Repo> {
505 
        let start = start.canonicalize()?;
506 
        let mut dir: Option<&Path> = Some(&start);
507 
        while let Some(d) = dir {
508 
            let oxy = d.join(OXY_DIR);
4e88e774 Andy McCall ↶ 509 
            if oxy.is_file() {
510 
                // A secondary workspace (spec §12.2): the file points at the repository.
511 
                let text = fs::read_to_string(&oxy)?;
512 
                let Some((shared, workspace)) = workspace::parse_workspace_file(&text) else {
513 
                    return fail(format!("{} is not a valid workspace file", oxy.display()));
514 
                };
515 
                let format = fs::read_to_string(shared.join("format"))?;
516 
                if format != FORMAT_LINE {
517 
                    return fail(format!("unsupported repository format: {}", format.trim()));
518 
                }
519 
                let config = Config::load(Some(&default_user_config()), &shared.join("config"))?;
d069701a Andy McCall ↶ 520 
                let store = open_objects(&shared, &config)?;
75167672 Andy McCall 521 
                let mut repo = Repo {
519c7c06 Andy McCall ↶ 522 
                    last_change: Default::default(),
98a1620a Andy McCall ↶ 523 
                    at_op: None,
1ec3f008 Andy McCall ↶ 524 
                    walked_dirs: Default::default(),
0389c3e8 Andy McCall ↶ 525 
                    listed_dirs: Default::default(),
75167672 Andy McCall 526 
                    git_export_skipped: Default::default(),
d0990c2b Andy McCall ↶ 527 
                    manifest_cache: Default::default(),
76740263 Andy McCall ↶ 528 
                    listing_cache: Default::default(),
d4dbab97 Andy McCall ↶ 529 
                    untracked_nested: Default::default(),
75167672 Andy McCall 530 
                    root: d.to_path_buf(),
531 
                    oxy: shared,
532 
                    store,
533 
                    workspace,
534 
                    bare: false,
535 
                    keys_dir: default_keys_dir(),
c863c7a7 Andy McCall ↶ 536 
                    platform: crate::portable::Platform::current(),
75167672 Andy McCall 537 
                    user_config: default_user_config(),
538 
                    config,
539 
                };
7b9809e6 Andy McCall ↶ 540 
                repo.apply_workspace_bookmark();
541 
                return Ok(repo);
4e88e774 Andy McCall ↶ 542 
            }
98286406 Andy McCall ↶ 543 
            if oxy.is_dir() {
544 
                let format = fs::read_to_string(oxy.join("format"))?;
545 
                let config = Config::load(Some(&default_user_config()), &oxy.join("config"))?;
e02bf0cd Andy McCall ↶ 546 
                let bare = oxy.join(BARE_MARKER).is_file();
547 
                let workspace = if bare {
548 
                    WorkspaceId::ZERO
549 
                } else {
75167672 Andy McCall 550 
                    WorkspaceId::from_hex(fs::read_to_string(oxy.join("workspace"))?.trim())
551 
                        .ok_or_else(|| RepoError::Msg("corrupt .oxy/workspace".into()))?
e02bf0cd Andy McCall ↶ 552 
                };
d069701a Andy McCall ↶ 553 
                let store = open_objects(&oxy, &config)?;
75167672 Andy McCall 554 
                let mut repo = Repo {
519c7c06 Andy McCall ↶ 555 
                    last_change: Default::default(),
98a1620a Andy McCall ↶ 556 
                    at_op: None,
1ec3f008 Andy McCall ↶ 557 
                    walked_dirs: Default::default(),
0389c3e8 Andy McCall ↶ 558 
                    listed_dirs: Default::default(),
75167672 Andy McCall 559 
                    git_export_skipped: Default::default(),
d0990c2b Andy McCall ↶ 560 
                    manifest_cache: Default::default(),
76740263 Andy McCall ↶ 561 
                    listing_cache: Default::default(),
d4dbab97 Andy McCall ↶ 562 
                    untracked_nested: Default::default(),
75167672 Andy McCall 563 
                    root: d.to_path_buf(),
564 
                    oxy,
565 
                    store,
566 
                    workspace,
567 
                    bare,
568 
                    keys_dir: default_keys_dir(),
c863c7a7 Andy McCall ↶ 569 
                    platform: crate::portable::Platform::current(),
75167672 Andy McCall 570 
                    user_config: default_user_config(),
571 
                    config,
572 
                };
e02bf0cd Andy McCall ↶ 573 
                if !bare {
574 
                    repo.apply_workspace_bookmark();
575 
                }
98286406 Andy McCall ↶ 576 
                if format == FORMAT1_LINE {
577 
                    repo.migrate_format1()?;
578 
                } else if format != FORMAT_LINE {
579 
                    return fail(format!("unsupported repository format: {}", format.trim()));
580 
                }
581 
                return Ok(repo);
582 
            }
583 
            dir = d.parent();
584 
        }
75167672 Andy McCall 585 
        fail(format!(
3e5772cc Andy McCall ↶ 586 
            "no oxy repository found at or above {}; start one with `oxy init` or `oxy clone URL`",
75167672 Andy McCall 587 
            start.display()
588 
        ))
98286406 Andy McCall ↶ 589 
    }
590
591 
    /// Rewrite a format-1 operation log, whose bookmarks were head lists,
592 
    /// into format 2, where each bookmark names a `state` object.
593 
    fn migrate_format1(&self) -> Result<()> {
dd21fd82 Andy McCall ↶ 594 
        let (head, version) = self.read_head()?;
98286406 Andy McCall ↶ 595 
        // Collect the chain root-first.
596 
        let mut chain: Vec<Format1Op> = Vec::new();
597 
        let mut cur = Some(head);
598 
        while let Some(r) = cur {
599 
            let op = Op::decode_format1(&self.store.get_typed(&r, ObjectType::Op)?)?;
600 
            cur = op.parents.first().copied();
601 
            chain.push(op);
602 
        }
603 
        chain.reverse();
604 
        let mut new_parent: Option<ObjectRef> = None;
605 
        for old in chain {
606 
            let mut view = View::default();
607 
            for (name, heads) in &old.view.bookmarks {
75167672 Andy McCall 608 
                view.bookmarks.insert(
609 
                    name.clone(),
610 
                    self.store.put_state(&StateObject::new(heads.clone()))?,
611 
                );
98286406 Andy McCall ↶ 612 
            }
613 
            view.workspaces = old.view.workspaces.clone();
75167672 Andy McCall 614 
            let op = Op {
615 
                parents: new_parent.into_iter().collect(),
616 
                hlc: old.hlc,
617 
                command: old.command,
618 
                view,
619 
            };
98286406 Andy McCall ↶ 620 
            new_parent = Some(self.store.put(ObjectType::Op, &op.encode())?);
621 
        }
dd21fd82 Andy McCall ↶ 622 
        if !self.swap_head(&new_parent.unwrap(), version.as_ref())? {
623 
            return fail("the operation log changed while it was rewritten; run the command again");
624 
        }
98286406 Andy McCall ↶ 625 
        write_atomic(&self.oxy.join("format"), FORMAT_LINE.as_bytes())?;
626 
        Ok(())
d02e545d Andy McCall ↶ 627 
    }
628
629 
    /// Pack automatically once many loose objects have accumulated.
630 
    pub(crate) fn maybe_autopack(&self) -> Result<()> {
631 
        if self.store.loose_count()? > AUTOPACK_THRESHOLD {
632 
            self.store.pack_loose()?;
135212de Andy McCall ↶ 633 
            self.store.merge_packs()?;
d02e545d Andy McCall ↶ 634 
        }
635 
        Ok(())
636 
    }
637
98286406 Andy McCall ↶ 638 
    /// The repository ID (spec §14.4). Repositories made before IDs existed get one on first use.
639 
    pub fn id(&self) -> Result<[u8; 16]> {
75167672 Andy McCall 640 
        if self.is_unbound() {
641 
            return Ok([0; 16]);
642 
        }
98286406 Andy McCall ↶ 643 
        let path = self.oxy.join("id");
644 
        match fs::read_to_string(&path) {
75167672 Andy McCall 645 
            Ok(h) => hex::decode(h.trim())
646 
                .ok()
647 
                .and_then(|v| v.try_into().ok())
648 
                .ok_or_else(|| RepoError::Msg("corrupt .oxy/id".into())),
98286406 Andy McCall ↶ 649 
            Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
650 
                let mut id = [0u8; 16];
651 
                rand::rngs::OsRng.fill_bytes(&mut id);
652 
                write_atomic(&path, hex::encode(id).as_bytes())?;
653 
                Ok(id)
654 
            }
655 
            Err(e) => Err(e.into()),
656 
        }
657 
    }
658
659 
    pub fn root(&self) -> &Path {
660 
        &self.root
661 
    }
662 
    pub fn store(&self) -> &Store {
663 
        &self.store
664 
    }
665 
    pub fn workspace(&self) -> WorkspaceId {
666 
        self.workspace
667 
    }
668
669 
    /// Use a different user directory for keys and the user config file
670 
    /// (tests, or `$OXY_HOME`). Reloads the effective configuration.
c863c7a7 Andy McCall ↶ 671 
    /// Act as if the working copy were on another platform's file system
672 
    /// (tests of Windows and macOS behaviour on any host).
673 
    pub fn with_platform(mut self, platform: crate::portable::Platform) -> Repo {
674 
        self.platform = platform;
675 
        self
676 
    }
677
678 
    pub fn platform(&self) -> crate::portable::Platform {
679 
        self.platform
680 
    }
681
98286406 Andy McCall ↶ 682 
    pub fn with_home(mut self, dir: impl Into<PathBuf>) -> Repo {
683 
        let dir = dir.into();
684 
        self.keys_dir = dir.join("keys");
685 
        self.user_config = dir.join("config.toml");
686 
        if let Ok(c) = Config::load(Some(&self.user_config), &self.oxy.join("config")) {
687 
            self.config = c;
688 
        }
7b9809e6 Andy McCall ↶ 689 
        self.apply_workspace_bookmark();
98286406 Andy McCall ↶ 690 
        self
691 
    }
692
7b9809e6 Andy McCall ↶ 693 
    /// The workspace's current bookmark (spec §10.1) overrides the
694 
    /// configured default when `oxy bookmark switch` has recorded one.
695 
    fn apply_workspace_bookmark(&mut self) {
75167672 Andy McCall 696 
        if let Ok(b) = fs::read_to_string(
697 
            self.oxy
698 
                .join("workspaces")
699 
                .join(self.workspace.to_hex())
700 
                .join("bookmark"),
701 
        ) {
7b9809e6 Andy McCall ↶ 702 
            let b = b.trim();
703 
            if !b.is_empty() {
704 
                self.config.bookmark = b.to_string();
705 
            }
706 
        }
707 
    }
708
98286406 Andy McCall ↶ 709 
    pub fn user_config_path(&self) -> &Path {
710 
        &self.user_config
711 
    }
712
713 
    /// Set a key in the user file and refresh the effective configuration.
714 
    pub fn set_user_config(&mut self, key: &str, value: &str) -> Result<()> {
715 
        Config::set_user(&self.user_config, key, value)?;
716 
        self.config = Config::load(Some(&self.user_config), &self.oxy.join("config"))?;
717 
        Ok(())
718 
    }
719
720 
    pub fn save_config(&self) -> Result<()> {
721 
        self.config.save(&self.oxy.join("config"))
722 
    }
723
724 
    fn author(&self) -> KeyId {
725 
        self.config.author_key.unwrap_or(ObjectRef::ZERO)
726 
    }
727 
}
728
729 
impl Repo {
730 
    fn new_working_copy(&self, deps: Vec<ChangeHash>, message: &str) -> Change {
731 
        let mut c = Change {
6f8aeb99 Andy McCall ↶ 732 
            version: crate::objects::CHANGE_VERSION,
98286406 Andy McCall ↶ 733 
            change_id: ChangeId::random(),
734 
            deps,
735 
            author: self.author(),
736 
            hlc: Hlc::now(),
737 
            message: message.into(),
738 
            tree_ops: vec![],
739 
            edits: vec![],
740 
            kind: ChangeKind::WorkingCopy,
741 
        };
742 
        c.normalize();
743 
        c
744 
    }
745
746 
    // ------------------------------------------------------------ operation log
747
748 
    pub fn head_op(&self) -> Result<(ObjectRef, Op)> {
98a1620a Andy McCall ↶ 749 
        if let Some(r) = self.at_op {
750 
            return Ok((r, Op::decode(&self.store.get_typed(&r, ObjectType::Op)?)?));
751 
        }
dd21fd82 Andy McCall ↶ 752 
        let r = self.head_ref()?;
98286406 Andy McCall ↶ 753 
        let op = Op::decode(&self.store.get_typed(&r, ObjectType::Op)?)?;
754 
        Ok((r, op))
755 
    }
756
757 
    pub fn view(&self) -> Result<View> {
758 
        Ok(self.head_op()?.1.view)
759 
    }
760
761 
    fn record_op(&self, parents: &[ObjectRef], command: &str, view: View) -> Result<ObjectRef> {
98a1620a Andy McCall ↶ 762 
        if self.at_op.is_some() {
763 
            return fail(
764 
                "--at-op shows the repository as it was; this command would change it (run it without --at-op)",
765 
            );
766 
        }
75167672 Andy McCall 767 
        let op = Op {
768 
            parents: parents.to_vec(),
769 
            hlc: Hlc::now(),
770 
            command: command.into(),
771 
            view,
772 
        };
98286406 Andy McCall ↶ 773 
        let r = self.store.put(ObjectType::Op, &op.encode())?;
4dcb8c09 Andy McCall ↶ 774 
        // The head is the commit point: everything it names must be on disk first.
236f35c0 Andy McCall ↶ 775 
        let flushing = crate::timing::phase("op: flush the store");
4dcb8c09 Andy McCall ↶ 776 
        self.store.flush()?;
236f35c0 Andy McCall ↶ 777 
        drop(flushing);
dd21fd82 Andy McCall ↶ 778 
        // Compare and swap (spec §12.1): when another process recorded an
779 
        // operation since this one read the head, the two are merged
780 
        // instead of one overwriting the other. On this disk the head moves
781 
        // under a short lock; on any other store by its record's version,
782 
        // and a record that moved meanwhile is read again and merged again.
783 
        let _lock = if self.store.is_local() {
784 
            Some(self.lock_ops()?)
785 
        } else {
786 
            None
787 
        };
788 
        let Some(parent) = parents.first() else {
19555025 Andy McCall ↶ 789 
            // The first operation (init): there is no head yet.
dd21fd82 Andy McCall ↶ 790 
            if !self.swap_head(&r, None)? {
791 
                return fail(format!(
792 
                    "{} already holds a repository's operation log",
793 
                    self.store.backend().describe()
794 
                ));
795 
            }
19555025 Andy McCall ↶ 796 
            return Ok(r);
797 
        };
dd21fd82 Andy McCall ↶ 798 
        let base = Op::decode(&self.store.get_typed(parent, ObjectType::Op)?)?.view;
799 
        for _ in 0..HEAD_TRIES {
800 
            let (current, version) = self.read_head()?;
801 
            let to = if current == *parent {
802 
                r
803 
            } else {
804 
                let theirs = Op::decode(&self.store.get_typed(&current, ObjectType::Op)?)?.view;
805 
                let merged = Op {
806 
                    parents: vec![current, r],
807 
                    hlc: Hlc::now(),
808 
                    command: format!("{command} (merged with a concurrent operation)"),
809 
                    view: self.merge_views(&base, &op.view, &theirs)?,
810 
                };
811 
                let m = self.store.put(ObjectType::Op, &merged.encode())?;
812 
                self.store.flush()?;
813 
                m
814 
            };
815 
            if self.swap_head(&to, version.as_ref())? {
816 
                return Ok(to);
817 
            }
818 
        }
819 
        fail("other oxy processes kept changing the operation log while this one recorded; run the command again")
19555025 Andy McCall ↶ 820 
    }
821
822 
    /// The operation head's lock, held only while the head pointer moves.
823 
    fn lock_ops(&self) -> Result<FileLock> {
824 
        fs::create_dir_all(self.oxy.join("ops"))?;
825 
        FileLock::acquire(
826 
            self.oxy.join("ops").join("lock"),
827 
            "another oxy process holds the operation log",
828 
        )
829 
    }
830
831 
    fn head_ref(&self) -> Result<ObjectRef> {
dd21fd82 Andy McCall ↶ 832 
        Ok(self.read_head()?.0)
833 
    }
834
835 
    /// The operation head, and on a store its record's version for
836 
    /// [`Repo::swap_head`]. On this disk the head is `.oxy/ops/head`; on
837 
    /// any other store (a bucket, or memory in tests) it is the store's
838 
    /// [`HEAD_RECORD`] reference record (oxygen#68).
839 
    fn read_head(&self) -> Result<(ObjectRef, Option<crate::strata::Version>)> {
840 
        if self.store.is_local() {
841 
            let hex = fs::read_to_string(self.oxy.join("ops").join("head"))?;
842 
            let r = ObjectRef::from_hex(hex.trim())
843 
                .ok_or_else(|| RepoError::Msg("corrupt .oxy/ops/head".into()))?;
844 
            return Ok((r, None));
845 
        }
846 
        let backend = self.store.backend();
847 
        match backend.read_ref(HEAD_RECORD) {
848 
            Ok(Some((bytes, version))) => {
849 
                let r = ObjectRef::from_hex(String::from_utf8_lossy(&bytes).trim()).ok_or_else(
850 
                    || {
851 
                        RepoError::Msg(format!(
852 
                            "the operation head in {} is corrupt",
853 
                            backend.describe()
854 
                        ))
855 
                    },
856 
                )?;
857 
                Ok((r, Some(version)))
858 
            }
859 
            Ok(None) => fail(format!(
860 
                "{} holds no operation log (no {HEAD_RECORD} record)",
861 
                backend.describe()
862 
            )),
863 
            Err(e) => fail(format!("read the operation head: {e}")),
864 
        }
19555025 Andy McCall ↶ 865 
    }
866
dd21fd82 Andy McCall ↶ 867 
    /// Move the operation head to `to` from the `expected` version
868 
    /// [`Repo::read_head`] gave (`None`: there is no head yet). On this
869 
    /// disk the caller holds the operations lock and the head always
870 
    /// moves; on a store it moves only if the record is unchanged, and
871 
    /// `false` says another process moved it first.
872 
    fn swap_head(&self, to: &ObjectRef, expected: Option<&crate::strata::Version>) -> Result<bool> {
873 
        if self.store.is_local() {
874 
            write_atomic(&self.oxy.join("ops").join("head"), to.to_hex().as_bytes())?;
875 
            return Ok(true);
876 
        }
877 
        match self
878 
            .store
879 
            .backend()
880 
            .write_ref(HEAD_RECORD, to.to_hex().as_bytes(), expected)
881 
        {
882 
            Ok(_) => Ok(true),
883 
            Err(crate::strata::Error::Conflict(_)) => Ok(false),
884 
            Err(e) => fail(format!("move the operation head: {e}")),
885 
        }
886 
    }
887
19555025 Andy McCall ↶ 888 
    /// Three-way merge of two views that both started from `base` (spec
889 
    /// §12.1). A reference only one side changed takes that side's value.
890 
    /// When both changed it differently: a bookmark keeps both sides'
6caacfff Andy McCall ↶ 891 
    /// heads (it may then have several), a change beats a deletion, a
892 
    /// workspace changed by both is an error (only one of them matches the
893 
    /// files on disk), and otherwise this process's value wins for remote
19555025 Andy McCall ↶ 894 
    /// references and the policy, the first value for a tag (tags never
895 
    /// move).
896 
    fn merge_views(&self, base: &View, ours: &View, theirs: &View) -> Result<View> {
897 
        fn pick<K: Ord + Clone, V: Clone + PartialEq>(
898 
            b: &BTreeMap<K, V>,
899 
            o: &BTreeMap<K, V>,
900 
            t: &BTreeMap<K, V>,
901 
            mut both: impl FnMut(&V, &V) -> Result<V>,
902 
        ) -> Result<BTreeMap<K, V>> {
903 
            let keys: BTreeSet<&K> = b.keys().chain(o.keys()).chain(t.keys()).collect();
904 
            let mut out = BTreeMap::new();
905 
            for k in keys {
906 
                let (bv, ov, tv) = (b.get(k), o.get(k), t.get(k));
907 
                let v = if ov == tv || tv == bv {
908 
                    ov.cloned()
909 
                } else if ov == bv {
910 
                    tv.cloned()
911 
                } else {
912 
                    match (ov, tv) {
913 
                        (Some(x), Some(y)) => Some(both(x, y)?),
914 
                        (Some(x), None) | (None, Some(x)) => Some(x.clone()),
915 
                        (None, None) => None,
916 
                    }
917 
                };
918 
                if let Some(v) = v {
919 
                    out.insert(k.clone(), v);
920 
                }
921 
            }
922 
            Ok(out)
923 
        }
924 
        let bookmarks = pick(
925 
            &base.bookmarks,
926 
            &ours.bookmarks,
927 
            &theirs.bookmarks,
928 
            |x, y| {
929 
                let mut heads = self.store.get_state(x)?.heads;
930 
                heads.extend(self.store.get_state(y)?.heads);
931 
                let heads = self.load_state(&heads)?.maximal(&heads);
932 
                Ok(self.store.put_state(&StateObject::new(heads))?)
933 
            },
934 
        )?;
6caacfff Andy McCall ↶ 935 
        // Two operations that both replaced one workspace's working copy
936 
        // cannot both be right: the files on disk match only one of them.
937 
        // The working-copy lock keeps this from happening; should it happen
938 
        // anyway, this operation is not recorded.
19555025 Andy McCall ↶ 939 
        let workspaces = pick(
940 
            &base.workspaces,
941 
            &ours.workspaces,
942 
            &theirs.workspaces,
6caacfff Andy McCall ↶ 943 
            |_, _| {
944 
                fail(
945 
                    "another oxy process changed this working copy at the same time; \
946 
                     this command was not recorded: check `oxy status` and run it again",
947 
                )
948 
            },
19555025 Andy McCall ↶ 949 
        )?;
950 
        let tags = pick(&base.tags, &ours.tags, &theirs.tags, |_, y| Ok(*y))?;
951 
        let empty = BTreeMap::new();
952 
        let names: BTreeSet<&String> = base
953 
            .remotes
954 
            .keys()
955 
            .chain(ours.remotes.keys())
956 
            .chain(theirs.remotes.keys())
957 
            .collect();
958 
        let mut remotes = BTreeMap::new();
959 
        for n in names {
960 
            let m = pick(
961 
                base.remotes.get(n).unwrap_or(&empty),
962 
                ours.remotes.get(n).unwrap_or(&empty),
963 
                theirs.remotes.get(n).unwrap_or(&empty),
964 
                |x, _| Ok(*x),
965 
            )?;
966 
            if !m.is_empty() {
967 
                remotes.insert(n.clone(), m);
968 
            }
969 
        }
970 
        let policy = if ours.policy == theirs.policy || theirs.policy == base.policy {
971 
            ours.policy
972 
        } else if ours.policy == base.policy {
973 
            theirs.policy
974 
        } else {
975 
            ours.policy.or(theirs.policy)
976 
        };
977 
        self.store.flush()?;
978 
        Ok(View {
979 
            bookmarks,
980 
            workspaces,
981 
            remotes,
982 
            policy,
983 
            tags,
984 
        })
98286406 Andy McCall ↶ 985 
    }
986
987 
    pub fn op_log(&self) -> Result<Vec<(ObjectRef, Op)>> {
988 
        let mut out = Vec::new();
989 
        let (mut r, mut op) = self.head_op()?;
990 
        loop {
991 
            let next = op.parents.first().copied();
992 
            out.push((r, op));
993 
            match next {
994 
                Some(p) => {
995 
                    r = p;
996 
                    op = Op::decode(&self.store.get_typed(&r, ObjectType::Op)?)?;
997 
                }
998 
                None => break,
999 
            }
1000 
        }
1001 
        Ok(out)
1002 
    }
1003
1004 
    /// Undo the last user command (spec §12.1). Operations that only recorded
1005 
    /// a working-copy snapshot are skipped, since the files are still on disk
98a1620a Andy McCall ↶ 1006 
    /// and the next snapshot recreates them, and a command's own checkout
1007 
    /// (`… (checkout)`) goes with the command. When the working copy is
1008 
    /// clean and was clean after the operation undone to, its files follow
1009 
    /// too, since nothing on disk can be lost; otherwise they stay as they
1010 
    /// are (an undone commit's edits come back as uncommitted work).
98286406 Andy McCall ↶ 1011 
    pub fn undo(&self) -> Result<Op> {
6caacfff Andy McCall ↶ 1012 
        let _wc = self.lock_working_copy()?;
98a1620a Andy McCall ↶ 1013 
        let clean_now =
1014 
            self.clean_by_manifest()? || self.snapshot_status("oxy undo")?.statuses.is_empty();
19555025 Andy McCall ↶ 1015 
        let (read_head, mut op) = self.head_op()?;
98a1620a Andy McCall ↶ 1016 
        let wc_now = op.view.workspaces.get(&self.workspace).copied();
1017 
        let up = |op: Op| -> Result<Option<Op>> {
1018 
            match op.parents.first() {
1019 
                Some(p) => Ok(Some(Op::decode(&self.store.get_typed(p, ObjectType::Op)?)?)),
1020 
                None => Ok(None),
1021 
            }
1022 
        };
1023 
        loop {
1024 
            while op.is_snapshot() {
1025 
                match up(op.clone())? {
1026 
                    Some(p) => op = p,
1027 
                    None => break,
1028 
                }
1029 
            }
1030 
            if !op.command.ends_with(" (checkout)") {
75167672 Andy McCall 1031 
                break;
98a1620a Andy McCall ↶ 1032 
            }
1033 
            match up(op.clone())? {
1034 
                Some(p) => op = p,
1035 
                None => break,
1036 
            }
98286406 Andy McCall ↶ 1037 
        }
1038 
        let Some(parent) = op.parents.first() else {
1039 
            return fail("nothing to undo");
1040 
        };
1041 
        let parent_op = Op::decode(&self.store.get_typed(parent, ObjectType::Op)?)?;
98a1620a Andy McCall ↶ 1042 
        {
dd21fd82 Andy McCall ↶ 1043 
            let _lock = if self.store.is_local() {
1044 
                Some(self.lock_ops()?)
1045 
            } else {
1046 
                None
1047 
            };
1048 
            let (now, version) = self.read_head()?;
1049 
            if now != read_head || !self.swap_head(parent, version.as_ref())? {
98a1620a Andy McCall ↶ 1050 
                return fail("another oxy command ran while undoing; run undo again");
1051 
            }
19555025 Andy McCall ↶ 1052 
        }
135212de Andy McCall ↶ 1053 
        self.drop_manifest();
98a1620a Andy McCall ↶ 1054 
        if let Some(then) = parent_op.view.workspaces.get(&self.workspace) {
1055 
            if clean_now && Some(*then) != wc_now {
1056 
                let wc = self.store.get_change(then)?;
1057 
                if wc.edits.is_empty() && wc.tree_ops.is_empty() {
1058 
                    let base = self.load_state_history(&wc.deps)?.maximal(&wc.deps);
1059 
                    // The files differ from the restored working copy by
1060 
                    // exactly what the undone command did: nothing else is
1061 
                    // on disk, as the working copy was clean.
1062 
                    self.checkout(&base, true, "oxy undo (checkout)")?;
1063 
                }
1064 
            }
1065 
        }
98286406 Andy McCall ↶ 1066 
        Ok(parent_op)
1067 
    }
1068
1069 
    /// The last operation that was not a snapshot.
1070 
    pub fn last_command(&self) -> Result<Op> {
1071 
        let (_, mut op) = self.head_op()?;
1072 
        while op.is_snapshot() {
75167672 Andy McCall 1073 
            let Some(parent) = op.parents.first() else {
1074 
                break;
1075 
            };
98286406 Andy McCall ↶ 1076 
            op = Op::decode(&self.store.get_typed(parent, ObjectType::Op)?)?;
1077 
        }
1078 
        Ok(op)
1079 
    }
1080
1081 
    // ------------------------------------------------------------ states
1082
252e8cb2 Andy McCall ↶ 1083 
    /// Load the dependency closure of `heads`, through a checkpoint when one applies.
98286406 Andy McCall ↶ 1084 
    pub fn load_state(&self, heads: &[ChangeHash]) -> Result<State> {
252e8cb2 Andy McCall ↶ 1085 
        if let Some(st) = self.load_state_checkpointed(heads)? {
1086 
            return Ok(st);
1087 
        }
1088 
        self.load_state_full(heads)
1089 
    }
1090
e8d38e47 Andy McCall ↶ 1091 
    /// The state of `heads` for its history (changes, tree, metadata) with
1092 
    /// no file graph built: much cheaper than [`Repo::load_state`] for log
1093 
    /// and friends. Build the graphs a command needs with
1094 
    /// [`Repo::load_graphs_of`].
1095 
    pub fn load_state_history(&self, heads: &[ChangeHash]) -> Result<State> {
1096 
        if let Some(st) = self.load_state_checkpointed_as(heads, true)? {
1097 
            return Ok(st);
1098 
        }
1099 
        self.load_state_replayed(heads, State::new_history_only())
1100 
    }
1101
1102 
    /// Build these file graphs in a history-only state.
1103 
    pub fn load_graphs_of(&self, state: &mut State, nodes: &[crate::ids::NodeId]) -> Result<()> {
5552e69c Andy McCall ↶ 1104 
        state
1105 
            .load_graphs(
1106 
                nodes,
1107 
                |refs| {
1108 
                    self.read_blobs(refs).map_err(|e| {
1109 
                        crate::objects::ObjectError::Schema(format!("graph and text blobs: {e}"))
1110 
                    })
1111 
                },
1112 
                |h| self.change_bytes(h),
1113 
            )
1114 
            .map_err(|e| self.index_damage(e))
cc7a0056 Andy McCall ↶ 1115 
    }
1116
5552e69c Andy McCall ↶ 1117 
    /// An error building graphs, and when it is a state index found
1118 
    /// damaged (a file's history that does not match its hash), the
1119 
    /// workspace's state indexes removed: they are rebuilt from history by
1120 
    /// the next command.
1121 
    fn index_damage(&self, e: crate::objects::ObjectError) -> RepoError {
1122 
        let crate::objects::ObjectError::DamagedIndex(why) = &e else {
1123 
            return e.into();
1124 
        };
1125 
        for path in self.state_indexes() {
1126 
            let _ = fs::remove_file(path);
1127 
        }
1128 
        RepoError::Msg(format!(
1129 
            "this workspace's state index was damaged ({why}) and has been removed; run the command again to rebuild it"
1130 
        ))
1131 
    }
1132
57d8dc05 Andy McCall ↶ 1133 
    /// The graphs of `nodes` of a history-only state, built beside it (see
1134 
    /// [`State::build_graphs_detached`]).
1135 
    pub fn build_graphs_detached(
1136 
        &self,
1137 
        state: &State,
1138 
        nodes: &[crate::ids::NodeId],
1139 
    ) -> Result<HashMap<crate::ids::NodeId, crate::graph::FileGraph>> {
1140 
        Ok(state
1141 
            .build_graphs_detached(
1142 
                nodes,
1143 
                |refs| {
1144 
                    self.read_blobs(refs).map_err(|e| {
1145 
                        crate::objects::ObjectError::Schema(format!("graph and text blobs: {e}"))
1146 
                    })
1147 
                },
1148 
                |h| self.change_bytes(h),
5552e69c Andy McCall ↶ 1149 
            )
1150 
            .map_err(|e| self.index_damage(e))?
57d8dc05 Andy McCall ↶ 1151 
            .into_iter()
1152 
            .collect())
1153 
    }
1154
091fef80 Andy McCall ↶ 1155 
    /// The contents of texts [`State::single_text`] gave: blobs read in
1156 
    /// one go, inline ones from their changes.
11e2b98a Andy McCall ↶ 1157 
    pub fn single_texts(&self, texts: &[crate::objects::TextRef]) -> Result<Vec<Vec<u8>>> {
091fef80 Andy McCall ↶ 1158 
        use crate::objects::TextRef;
1159 
        let refs: Vec<ObjectRef> = texts
1160 
            .iter()
1161 
            .filter_map(|t| match t {
1162 
                TextRef::Blob(r, _, _) => Some(*r),
1163 
                TextRef::Inline { .. } => None,
1164 
            })
1165 
            .collect();
1166 
        let mut blobs = self.read_blobs(&refs)?.into_iter();
1167 
        let mut out = Vec::with_capacity(texts.len());
1168 
        for t in texts {
1169 
            let lines = match t {
1170 
                TextRef::Blob(_, count, form) => {
1171 
                    let content = blobs
1172 
                        .next()
1173 
                        .ok_or_else(|| RepoError::Msg("text blobs: wrong count".into()))?;
1174 
                    crate::objects::text_lines(&content, *count, *form)?
1175 
                }
1176 
                TextRef::Inline { change, start, end } => {
1177 
                    let bytes = self.change_bytes(change)?;
1178 
                    let range = bytes.get(*start as usize..*end as usize).ok_or_else(|| {
1179 
                        RepoError::Msg("an inline insert lies outside its change".into())
1180 
                    })?;
1181 
                    crate::objects::inline_lines(range)?
1182 
                }
1183 
            };
1184 
            out.push(lines.bytes().to_vec());
1185 
        }
1186 
        Ok(out)
1187 
    }
1188
cc7a0056 Andy McCall ↶ 1189 
    /// Build every graph a history-only state still lacks, making it full.
1190 
    pub fn load_all_graphs(&self, state: &mut State) -> Result<()> {
5552e69c Andy McCall ↶ 1191 
        state
1192 
            .load_all_graphs(
1193 
                |refs| {
1194 
                    self.read_blobs(refs).map_err(|e| {
1195 
                        crate::objects::ObjectError::Schema(format!("graph and text blobs: {e}"))
1196 
                    })
1197 
                },
1198 
                |h| self.change_bytes(h),
1199 
            )
1200 
            .map_err(|e| self.index_damage(e))
a4fad5b0 Andy McCall ↶ 1201 
    }
1202
519c7c06 Andy McCall ↶ 1203 
    /// A change's encoding, for building graphs of its inline inserts. The
acadb5ed Andy McCall ↶ 1204 
    /// last few read are kept: a checkout building graphs a batch at a
1205 
    /// time reads the same large changes for every batch, often in turn.
a4fad5b0 Andy McCall ↶ 1206 
    fn change_bytes(
1207 
        &self,
1208 
        h: &ChangeHash,
519c7c06 Andy McCall ↶ 1209 
    ) -> std::result::Result<std::sync::Arc<Vec<u8>>, crate::objects::ObjectError> {
acadb5ed Andy McCall ↶ 1210 
        let mut cache = self.last_change.borrow_mut();
1211 
        if let Some(i) = cache.iter().position(|(k, _)| k == h) {
1212 
            let hit = cache.remove(i);
1213 
            let b = hit.1.clone();
1214 
            cache.push(hit);
1215 
            return Ok(b);
519c7c06 Andy McCall ↶ 1216 
        }
1217 
        let b = std::sync::Arc::new(self.store.get_typed(h, ObjectType::Change).map_err(|e| {
a4fad5b0 Andy McCall ↶ 1218 
            crate::objects::ObjectError::Schema(format!("change {}: {e}", h.short(12)))
519c7c06 Andy McCall ↶ 1219 
        })?);
acadb5ed Andy McCall ↶ 1220 
        cache.push((*h, b.clone()));
1221 
        // The last few, within a memory budget; the newest always stays.
1222 
        while cache.len() > CHANGE_CACHE_COUNT
1223 
            || (cache.len() > 1
1224 
                && cache.iter().map(|(_, b)| b.len()).sum::<usize>() > CHANGE_CACHE_BYTES)
1225 
        {
1226 
            cache.remove(0);
1227 
        }
519c7c06 Andy McCall ↶ 1228 
        Ok(b)
e8d38e47 Andy McCall ↶ 1229 
    }
1230
252e8cb2 Andy McCall ↶ 1231 
    /// Load the dependency closure of `heads` change by change.
1232 
    pub fn load_state_full(&self, heads: &[ChangeHash]) -> Result<State> {
e8d38e47 Andy McCall ↶ 1233 
        self.load_state_replayed(heads, State::new())
1234 
    }
1235
1236 
    fn load_state_replayed(&self, heads: &[ChangeHash], mut state: State) -> Result<State> {
86e1824f Andy McCall ↶ 1237 
        let history = state.is_history_only();
98286406 Andy McCall ↶ 1238 
        let mut stack: Vec<ChangeHash> = heads.to_vec();
1239 
        let mut seen: BTreeSet<ChangeHash> = BTreeSet::new();
0bdc3f91 Andy McCall ↶ 1240 
        {
1241 
            let _t = crate::timing::phase("load: read and add changes");
d0d52107 Andy McCall ↶ 1242 
            let (mut read, mut add, mut fetch) = (
1243 
                std::time::Duration::ZERO,
1244 
                std::time::Duration::ZERO,
1245 
                std::time::Duration::ZERO,
1246 
            );
0bdc3f91 Andy McCall ↶ 1247 
            while let Some(h) = stack.pop() {
f9466021 Andy McCall ↶ 1248 
                // A change the state holds brings its history with it.
1249 
                if !seen.insert(h) || state.contains(&h) {
0bdc3f91 Andy McCall ↶ 1250 
                    continue;
1251 
                }
1252 
                let t0 = std::time::Instant::now();
86e1824f Andy McCall ↶ 1253 
                // A history-only state reads text only for the graphs it
1254 
                // builds; either way the hash is the one asked for.
1255 
                let (c, texts) = if history {
d0d52107 Andy McCall ↶ 1256 
                    let bytes = self.store.get_typed(&h, ObjectType::Change)?;
1257 
                    fetch += t0.elapsed();
1258 
                    Change::decode_skeleton(&bytes, h)
1259 
                        .map_err(|e| crate::store::StoreError::Corrupt(h, e.to_string()))?
86e1824f Andy McCall ↶ 1260 
                } else {
1261 
                    (self.store.get_change(&h)?, Vec::new())
1262 
                };
0bdc3f91 Andy McCall ↶ 1263 
                read += t0.elapsed();
1264 
                stack.extend(c.deps.iter().copied());
1265 
                let t1 = std::time::Instant::now();
86e1824f Andy McCall ↶ 1266 
                state.add_known(h, c, texts);
0bdc3f91 Andy McCall ↶ 1267 
                add += t1.elapsed();
98286406 Andy McCall ↶ 1268 
            }
0bdc3f91 Andy McCall ↶ 1269 
            crate::timing::record("load:   of which reading", read);
d0d52107 Andy McCall ↶ 1270 
            crate::timing::record("load:     of which fetching", fetch);
0bdc3f91 Andy McCall ↶ 1271 
            crate::timing::record("load:   of which applying", add);
98286406 Andy McCall ↶ 1272 
        }
0bdc3f91 Andy McCall ↶ 1273 
        let _t = crate::timing::phase("load: validate");
98286406 Andy McCall ↶ 1274 
        state.validate()?;
1275 
        Ok(state)
1276 
    }
1277
1278 
    pub fn working_copy(&self) -> Result<(ChangeHash, Change)> {
e02bf0cd Andy McCall ↶ 1279 
        if self.bare {
1280 
            return fail("this is a bare repository: it has no working copy");
1281 
        }
98286406 Andy McCall ↶ 1282 
        let view = self.view()?;
1283 
        let h = *view
1284 
            .workspaces
1285 
            .get(&self.workspace)
1286 
            .ok_or_else(|| RepoError::Msg("this workspace is not in the current view".into()))?;
1287 
        Ok((h, self.store.get_change(&h)?))
1288 
    }
1289
1290 
    /// Heads of a bookmark's state (spec §10.1).
1291 
    pub fn heads_of(&self, view: &View, bookmark: &str) -> Result<Vec<ChangeHash>> {
1292 
        match view.bookmarks.get(bookmark) {
1293 
            Some(st) => Ok(self.store.get_state(st)?.heads),
1294 
            None => Ok(vec![]),
1295 
        }
1296 
    }
1297
1298 
    pub fn bookmark_heads(&self) -> Result<Vec<ChangeHash>> {
1299 
        self.heads_of(&self.view()?, &self.config.bookmark)
1300 
    }
1301
1302 
    // ------------------------------------------------------------ blobs
1303
1304 
    fn blob_for(content: &[u8]) -> Blob {
6f8aeb99 Andy McCall ↶ 1305 
        Blob::of(content)
98286406 Andy McCall ↶ 1306 
    }
1307
1308 
    pub fn blob_ref(content: &[u8]) -> ObjectRef {
1309 
        Self::blob_for(content).hash()
1310 
    }
1311
1312 
    pub fn store_blob(&self, content: &[u8]) -> Result<ObjectRef> {
1313 
        for c in crate::chunk::chunks(content) {
1314 
            self.store.put_chunk(c)?;
1315 
        }
75167672 Andy McCall 1316 
        Ok(self
1317 
            .store
1318 
            .put(ObjectType::Blob, &Self::blob_for(content).encode())?)
98286406 Andy McCall ↶ 1319 
    }
1320
afaba5f1 Andy McCall ↶ 1321 
    /// Read many blobs at once (checkpoint graphs): the blob objects, then
1322 
    /// all their chunks, are read in parallel; a blob the store lacks goes
1323 
    /// through [`Repo::read_blob`], which may fetch it lazily.
1324 
    pub fn read_blobs(&self, refs: &[ObjectRef]) -> Result<Vec<Vec<u8>>> {
1325 
        let mut blobs: Vec<Option<Blob>> = Vec::with_capacity(refs.len());
1326 
        for (r, got) in refs.iter().zip(self.store.get_many(refs)) {
1327 
            blobs.push(match got {
1328 
                Ok((ObjectType::Blob, body)) => Some(Blob::decode(&body)?),
1329 
                Ok((ty, _)) => return fail(format!("{} is a {ty:?}, not a blob", r.short(12))),
1330 
                Err(crate::store::StoreError::NotFound(_)) => None,
1331 
                Err(e) => return Err(e.into()),
1332 
            });
1333 
        }
1334 
        let chunk_refs: Vec<ObjectRef> = blobs
1335 
            .iter()
1336 
            .flatten()
1337 
            .flat_map(|b| b.chunks.iter().copied())
1338 
            .collect();
1339 
        let mut chunks = self.store.get_many(&chunk_refs).into_iter();
1340 
        let mut out = Vec::with_capacity(refs.len());
1341 
        for (r, blob) in refs.iter().zip(blobs) {
1342 
            match blob {
1343 
                None => out.push(self.read_blob(r)?),
1344 
                Some(b) => {
1345 
                    let mut content = Vec::with_capacity(b.length as usize);
1346 
                    for c in &b.chunks {
1347 
                        match chunks.next() {
1348 
                            Some(Ok((ObjectType::Chunk, body))) => content.extend(body),
1349 
                            Some(Ok(_)) => return fail(format!("{} is not a chunk", c.short(12))),
1350 
                            Some(Err(e)) => return Err(e.into()),
1351 
                            None => return fail("chunk list ended early"),
1352 
                        }
1353 
                    }
1354 
                    out.push(content);
1355 
                }
1356 
            }
1357 
        }
1358 
        Ok(out)
1359 
    }
1360
b3c5a59a Andy McCall ↶ 1361 
    /// Read a blob's content, fetching it from the remote that withheld it
1362 
    /// when it is lazily absent (spec §13.5).
98286406 Andy McCall ↶ 1363 
    pub fn read_blob(&self, r: &ObjectRef) -> Result<Vec<u8>> {
b3c5a59a Andy McCall ↶ 1364 
        if !self.store.has(r) && self.lazy_remote(r).is_some() {
1365 
            self.fetch_lazy_blob(r)?;
1366 
        }
98286406 Andy McCall ↶ 1367 
        let blob = Blob::decode(&self.store.get_typed(r, ObjectType::Blob)?)?;
1368 
        let mut out = Vec::with_capacity(blob.length as usize);
1369 
        for c in &blob.chunks {
1370 
            out.extend(self.store.get_typed(c, ObjectType::Chunk)?);
1371 
        }
1372 
        Ok(out)
1373 
    }
1374
1375 
    // ------------------------------------------------------------ working copy
1376
cdb5beba Andy McCall ↶ 1377 
    /// Attributes from the `.oxyattributes` files in the working directory.
1378 
    pub fn disk_attributes(&self) -> crate::attributes::Attributes {
f9466021 Andy McCall ↶ 1379 
        // Nested repositories and ignored paths are left out, as in a walk.
1380 
        let listing = walk::list(&self.root, &|p: &str| is_attributes(p)).unwrap_or_default();
1381 
        let files: Vec<(String, Vec<u8>)> = listing
1382 
            .files
1383 
            .into_iter()
1384 
            .filter_map(|(rel, path, _)| Some((rel, fs::read(path).ok()?)))
1385 
            .collect();
75167672 Andy McCall 1386 
        crate::attributes::Attributes::from_files(
1387 
            files.iter().map(|(p, c)| (p.as_str(), c.as_slice())),
1388 
        )
cdb5beba Andy McCall ↶ 1389 
    }
1390
d4dbab97 Andy McCall ↶ 1391 
    /// Folders holding a repository of their own that the last look at the
1392 
    /// working copy left out, not being subrepos (oxygen#43).
1393 
    pub fn untracked_nested(&self) -> Vec<String> {
1394 
        self.untracked_nested.borrow().clone()
1395 
    }
1396
cdb5beba Andy McCall ↶ 1397 
    /// Attributes from a materialized state's `.oxyattributes` files.
1398 
    pub fn state_attributes(snapshot: &Snapshot) -> crate::attributes::Attributes {
75167672 Andy McCall 1399 
        crate::attributes::Attributes::from_files(snapshot.files.iter().filter_map(|(p, f)| {
1400 
            match &f.content {
1401 
                FileContent::Text { bytes, .. }
1402 
                    if p == ".oxyattributes" || p.ends_with("/.oxyattributes") =>
1403 
                {
1404 
                    Some((p.as_str(), bytes.as_slice()))
1405 
                }
1406 
                _ => None,
1407 
            }
cdb5beba Andy McCall ↶ 1408 
        }))
1409 
    }
1410
1411 
    /// Store a blob, raw when the path is `nocompress`.
1412 
    pub fn store_blob_for(&self, content: &[u8], nocompress: bool) -> Result<ObjectRef> {
1413 
        if !nocompress {
1414 
            return self.store_blob(content);
1415 
        }
1416 
        for c in crate::chunk::chunks(content) {
1417 
            self.store.put_chunk_nocompress(c)?;
1418 
        }
75167672 Andy McCall 1419 
        Ok(self
1420 
            .store
1421 
            .put(ObjectType::Blob, &Self::blob_for(content).encode())?)
cdb5beba Andy McCall ↶ 1422 
    }
1423
f9575918 Andy McCall ↶ 1424 
    /// This workspace's state index (oxygen#41): its working copy's base,
1425 
    /// saved so a snapshot reads it rather than replaying history.
638fb214 Andy McCall ↶ 1426 
    /// The state index of the state with these heads: this workspace keeps
1427 
    /// the last few it used, named by their heads, so moving between two
1428 
    /// states finds both.
1429 
    fn state_index_path(&self, heads: &[ChangeHash]) -> PathBuf {
1430 
        let mut heads = heads.to_vec();
1431 
        heads.sort();
1432 
        let mut h = blake3::Hasher::new();
1433 
        for c in &heads {
1434 
            h.update(&c.0);
1435 
        }
1436 
        self.oxy.join("index").join(format!(
1437 
            "{}{}",
1438 
            self.state_index_prefix(),
1439 
            &h.finalize().to_hex()[..16]
1440 
        ))
f9575918 Andy McCall ↶ 1441 
    }
1442
638fb214 Andy McCall ↶ 1443 
    fn state_index_prefix(&self) -> String {
1444 
        format!("state-{}-", self.workspace.to_hex())
1445 
    }
1446
1447 
    /// This workspace's state indexes, newest first.
1448 
    fn state_indexes(&self) -> Vec<PathBuf> {
1449 
        let prefix = self.state_index_prefix();
1450 
        let mut found: Vec<(std::time::SystemTime, PathBuf)> = fs::read_dir(self.oxy.join("index"))
1451 
            .into_iter()
1452 
            .flatten()
1453 
            .flatten()
1454 
            .filter(|e| e.file_name().to_string_lossy().starts_with(&prefix))
1455 
            .filter_map(|e| Some((e.metadata().ok()?.modified().ok()?, e.path())))
1456 
            .collect();
1457 
        found.sort_by_key(|f| std::cmp::Reverse(f.0));
1458 
        found.into_iter().map(|(_, p)| p).collect()
1459 
    }
1460
f9575918 Andy McCall ↶ 1461 
    /// The history-only state of `deps` (as [`Repo::load_state_history`]),
1462 
    /// from this workspace's state index when it holds that history: every
1463 
    /// one of `deps` is in it and their heads are its heads. Otherwise it
1464 
    /// is loaded from history and indexed for next time.
1465 
    pub fn load_state_indexed(&self, deps: &[ChangeHash]) -> Result<State> {
638fb214 Andy McCall ↶ 1466 
        if let Some((st, heads)) = self.read_state_index(deps) {
f9466021 Andy McCall ↶ 1467 
            let mut have = st.maximal(deps);
1468 
            have.sort();
1469 
            if deps.iter().all(|d| st.contains(d)) && have == heads {
5552e69c Andy McCall ↶ 1470 
                // Re-encoding reads every history: one found damaged is
1471 
                // left to the build that reads it, which reports it.
f9466021 Andy McCall ↶ 1472 
                #[cfg(test)]
5552e69c Andy McCall ↶ 1473 
                if let Some(mine) = st.write_index() {
1474 
                    assert_eq!(
1475 
                        Some(mine),
1476 
                        self.load_state_history(deps)?.write_index(),
1477 
                        "a state index holds the state history gives"
1478 
                    );
1479 
                }
f9466021 Andy McCall ↶ 1480 
                return Ok(st);
1481 
            }
1482 
            // An index a few changes behind (a commit, a pull since): add
1483 
            // what it lacks, as long as all it holds is history of `deps`.
1484 
            // Anything amiss (a change beyond a shallow boundary) takes the
1485 
            // long way instead.
1486 
            let (indexed, started) = (st.len(), std::time::Instant::now());
1487 
            if let Ok(st) = self.load_state_replayed(deps, st) {
1488 
                let mut all = deps.to_vec();
1489 
                all.extend(heads.iter().copied());
1490 
                let (mut within, mut want) = (st.maximal(&all), st.maximal(deps));
1491 
                within.sort();
1492 
                want.sort();
1493 
                if within == want {
1494 
                    #[cfg(test)]
1495 
                    self.assert_same_state(&st, &self.load_state_history(deps)?);
1496 
                    // Writing the index costs far more than replaying a
1497 
                    // few small changes: it is rewritten once it falls
1498 
                    // behind, or when replaying took long (large changes).
1499 
                    if st.len() - indexed > INDEX_REPLAY_LIMIT
1500 
                        || started.elapsed() > INDEX_REPLAY_TIME
1501 
                    {
1502 
                        self.write_state_index(deps, &st);
1503 
                    }
1504 
                    return Ok(st);
1505 
                }
1506 
            }
f9575918 Andy McCall ↶ 1507 
        }
1508 
        let st = self.load_state_history(deps)?;
1509 
        self.write_state_index(deps, &st);
1510 
        Ok(st)
f9466021 Andy McCall ↶ 1511 
    }
1512
1513 
    /// Two history-only states hold the same changes and render the same
1514 
    /// files. Their indexes may differ: one loaded from a checkpoint keeps
1515 
    /// the checkpoint's graphs as the bases of its files' histories.
1516 
    #[cfg(test)]
1517 
    fn assert_same_state(&self, a: &State, b: &State) {
1518 
        let ids = |s: &State| s.changes().map(|(h, _)| *h).collect::<BTreeSet<_>>();
1519 
        assert_eq!(ids(a), ids(b), "the same changes");
1520 
        let files = |s: &State| {
1521 
            let mut s = s.clone();
1522 
            self.load_all_graphs(&mut s).unwrap();
1523 
            s.materialize().files
1524 
        };
1525 
        assert_eq!(files(a), files(b), "the same files");
f9575918 Andy McCall ↶ 1526 
    }
1527
4cce912a Andy McCall ↶ 1528 
    /// The state index file: its heads, a hash of its changes block and one
1529 
    /// of the rest, then the body [`State::write_index`] gives, whose first
1530 
    /// block is the changes. Hashed in two parts so the log can read and
1531 
    /// check the changes alone.
638fb214 Andy McCall ↶ 1532 
    /// The index of `deps` if there is one, else the newest, which the
1533 
    /// caller may bring up to date.
1534 
    fn read_state_index(&self, deps: &[ChangeHash]) -> Option<(State, Vec<ChangeHash>)> {
5552e69c Andy McCall ↶ 1535 
        use std::io::Read;
f9575918 Andy McCall ↶ 1536 
        let _t = crate::timing::phase("load: state index");
638fb214 Andy McCall ↶ 1537 
        let exact = self.state_index_path(deps);
5552e69c Andy McCall ↶ 1538 
        let reading = crate::timing::phase("load:   of which reading the file");
1539 
        let mut f = match fs::File::open(&exact) {
1540 
            Ok(f) => {
638fb214 Andy McCall ↶ 1541 
                // Used now: the newest, so the last to be dropped.
1542 
                let _ = fs::File::options()
1543 
                    .write(true)
1544 
                    .open(&exact)
1545 
                    .and_then(|f| f.set_modified(std::time::SystemTime::now()));
5552e69c Andy McCall ↶ 1546 
                f
638fb214 Andy McCall ↶ 1547 
            }
5552e69c Andy McCall ↶ 1548 
            Err(_) => fs::File::open(self.state_indexes().first()?).ok()?,
638fb214 Andy McCall ↶ 1549 
        };
5552e69c Andy McCall ↶ 1550 
        let total = f.metadata().ok()?.len();
1551 
        let mut word = [0u8; 4];
1552 
        f.read_exact(&mut word).ok()?;
1553 
        let n = u32::from_le_bytes(word) as usize;
1554 
        let mut head = vec![0u8; 32 * n + 64 + 8];
1555 
        f.read_exact(&mut head).ok()?;
1556 
        let heads: Vec<ChangeHash> = head[..32 * n]
1557 
            .chunks_exact(32)
1558 
            .map(|c| ObjectRef(c.try_into().unwrap()))
1559 
            .collect();
1560 
        let (changes_sum, rest_sum) = (&head[32 * n..32 * n + 32], &head[32 * n + 32..32 * n + 64]);
1561 
        let front_len = u64::from_le_bytes(head[32 * n + 64..].try_into().unwrap());
1562 
        let at = (4 + head.len()) as u64;
1563 
        if at.checked_add(front_len).is_none_or(|end| end > total) {
1564 
            return None;
f9575918 Andy McCall ↶ 1565 
        }
5552e69c Andy McCall ↶ 1566 
        // The front only: the files' histories after it are read, each
1567 
        // checked by its own hash, when a command builds that file.
1568 
        let mut front = vec![0u8; front_len as usize];
1569 
        f.read_exact(&mut front).ok()?;
1570 
        drop(reading);
1571 
        let checking = crate::timing::phase("load:   of which checking the file");
1572 
        let split = 4 + u32::from_le_bytes(front.get(..4)?.try_into().unwrap()) as usize;
1573 
        let (changes, rest) = (front.get(..split)?, front.get(split..)?);
79b5a90f Andy McCall ↶ 1574 
        let sum = |b: &[u8]| *blake3::Hasher::new().update_rayon(b).finalize().as_bytes();
5552e69c Andy McCall ↶ 1575 
        if sum(changes) != changes_sum || sum(rest) != rest_sum {
f9575918 Andy McCall ↶ 1576 
            return None;
1577 
        }
5552e69c Andy McCall ↶ 1578 
        drop(checking);
1579 
        let histories = crate::state::Histories::File(
1580 
            std::sync::Arc::new(f),
1581 
            at + front_len,
1582 
            total - at - front_len,
1583 
        );
1584 
        Some((State::read_index_front(&front, histories).ok()?, heads))
4cce912a Andy McCall ↶ 1585 
    }
1586
1587 
    /// The changes of this workspace's state index (the closure of its
1588 
    /// heads), read without the rest of the index.
bf0247ee Andy McCall ↶ 1589 
    #[allow(clippy::type_complexity)]
4cce912a Andy McCall ↶ 1590 
    pub(crate) fn read_state_index_changes(
1591 
        &self,
bf0247ee Andy McCall ↶ 1592 
    ) -> Option<(
1593 
        Vec<ChangeHash>,
1594 
        crate::fasthash::FastMap<ChangeHash, std::sync::Arc<crate::state::ChangeMeta>>,
1595 
    )> {
4cce912a Andy McCall ↶ 1596 
        use std::io::Read;
1597 
        let _t = crate::timing::phase("load: state index changes");
638fb214 Andy McCall ↶ 1598 
        let mut f = fs::File::open(self.state_indexes().first()?).ok()?;
4cce912a Andy McCall ↶ 1599 
        let mut word = [0u8; 4];
1600 
        f.read_exact(&mut word).ok()?;
1601 
        let n = u32::from_le_bytes(word) as usize;
5552e69c Andy McCall ↶ 1602 
        // Heads, the two sums, the front's length, then the changes block.
1603 
        let mut head = vec![0u8; 32 * n + 64 + 8 + 4];
4cce912a Andy McCall ↶ 1604 
        f.read_exact(&mut head).ok()?;
bf0247ee Andy McCall ↶ 1605 
        let heads: Vec<ChangeHash> = head[..32 * n]
1606 
            .chunks_exact(32)
1607 
            .map(|c| ObjectRef(c.try_into().unwrap()))
1608 
            .collect();
4cce912a Andy McCall ↶ 1609 
        let sum = &head[32 * n..32 * n + 32];
5552e69c Andy McCall ↶ 1610 
        let len = u32::from_le_bytes(head[32 * n + 72..].try_into().unwrap()) as usize;
4cce912a Andy McCall ↶ 1611 
        let mut changes = vec![0u8; 4 + len];
5552e69c Andy McCall ↶ 1612 
        changes[..4].copy_from_slice(&head[32 * n + 72..]);
4cce912a Andy McCall ↶ 1613 
        f.read_exact(&mut changes[4..]).ok()?;
79b5a90f Andy McCall ↶ 1614 
        let checking = crate::timing::phase("load:   of which checking");
1615 
        let mut h = blake3::Hasher::new();
1616 
        h.update_rayon(&changes);
1617 
        if h.finalize().as_bytes() != sum {
4cce912a Andy McCall ↶ 1618 
            return None;
1619 
        }
79b5a90f Andy McCall ↶ 1620 
        drop(checking);
1621 
        let _decoding = crate::timing::phase("load:   of which decoding");
bf0247ee Andy McCall ↶ 1622 
        Some((heads, State::read_index_changes(&changes[4..]).ok()?))
f9575918 Andy McCall ↶ 1623 
    }
1624
1625 
    /// Save `st`, the history-only state of `deps`, as this workspace's
1626 
    /// state index; quietly nothing when it cannot be (it is only a cache).
1627 
    pub(crate) fn write_state_index(&self, deps: &[ChangeHash], st: &State) {
1628 
        let _t = crate::timing::phase("write: state index");
f9466021 Andy McCall ↶ 1629 
        let encoding = crate::timing::phase("write:   of which encoding");
f9575918 Andy McCall ↶ 1630 
        let Some(body) = st.write_index() else {
1631 
            return;
1632 
        };
f9466021 Andy McCall ↶ 1633 
        drop(encoding);
3f1b8fe2 Andy McCall ↶ 1634 
        #[cfg(test)]
1635 
        assert_eq!(
1636 
            State::read_index(&body).ok().and_then(|s| s.write_index()),
1637 
            Some(body.clone()),
1638 
            "a state index reads back as written"
1639 
        );
f9575918 Andy McCall ↶ 1640 
        let mut heads = st.maximal(deps);
1641 
        heads.sort();
1642 
        let mut w = crate::flat::Writer::default();
1643 
        w.count(heads.len());
1644 
        for h in &heads {
1645 
            w.object(h);
1646 
        }
5552e69c Andy McCall ↶ 1647 
        // The body is the front's length, the front, then the files'
1648 
        // histories, which carry their own hashes: the sums cover the
1649 
        // front's changes block and the rest of the front.
1650 
        let front_len = u64::from_le_bytes(body[..8].try_into().unwrap()) as usize;
1651 
        let front = &body[8..8 + front_len];
1652 
        let split = 4 + u32::from_le_bytes(front[..4].try_into().unwrap()) as usize;
1653 
        let sum = |b: &[u8]| *blake3::Hasher::new().update_rayon(b).finalize().as_bytes();
1654 
        w.fixed(&sum(&front[..split]));
1655 
        w.fixed(&sum(&front[split..]));
f9575918 Andy McCall ↶ 1656 
        w.fixed(&body);
638fb214 Andy McCall ↶ 1657 
        let path = self.state_index_path(&heads);
f9466021 Andy McCall ↶ 1658 
        let _ = fs::create_dir_all(self.oxy.join("index"));
1659 
        let _ = write_cache(&path, &w.0);
638fb214 Andy McCall ↶ 1660 
        // The last two used: the index of a state moved away from stays for
1661 
        // a move back. Older ones (and the unnamed one of before) go.
1662 
        for old in self.state_indexes().iter().skip(STATE_INDEXES_KEPT) {
1663 
            let _ = fs::remove_file(old);
1664 
        }
1665 
        let _ = fs::remove_file(
1666 
            self.oxy
1667 
                .join("index")
1668 
                .join(format!("state-{}", self.workspace.to_hex())),
1669 
        );
f9575918 Andy McCall ↶ 1670 
    }
1671
cdb5beba Andy McCall ↶ 1672 
    fn locks_dir(&self) -> PathBuf {
1673 
        self.oxy.join("locks")
1674 
    }
1675
1676 
    /// Whether `path` is locked locally for editing (spec §18, advisory).
1677 
    pub fn is_locked(&self, path: &str) -> bool {
1678 
        self.locks_dir().join(path.replace('/', "%2F")).exists()
1679 
    }
1680
1681 
    /// Take the advisory local lock on a `lockable` file and make it writable.
1682 
    pub fn lock_path(&self, path: &str) -> Result<()> {
1683 
        let attrs = self.disk_attributes();
1684 
        if !attrs.lookup(path).lockable {
1685 
            return fail(format!("{path} is not marked lockable in .oxyattributes"));
1686 
        }
1687 
        fs::create_dir_all(self.locks_dir())?;
75167672 Andy McCall 1688 
        write_atomic(
1689 
            &self.locks_dir().join(path.replace('/', "%2F")),
1690 
            std::process::id().to_string().as_bytes(),
1691 
        )?;
cdb5beba Andy McCall ↶ 1692 
        set_readonly(&self.root.join(path), false)?;
1693 
        Ok(())
1694 
    }
1695
1696 
    /// Release the local lock and make the file read-only again.
1697 
    pub fn unlock_path(&self, path: &str) -> Result<()> {
1698 
        let _ = fs::remove_file(self.locks_dir().join(path.replace('/', "%2F")));
1699 
        if self.root.join(path).exists() {
1700 
            set_readonly(&self.root.join(path), true)?;
1701 
        }
1702 
        Ok(())
1703 
    }
1704
7421309f Andy McCall ↶ 1705 
    /// Whether a repository path is inside the sparse filter (spec §17).
1706 
    pub fn in_scope(&self, path: &str) -> bool {
75167672 Andy McCall 1707 
        self.config.sparse_paths.is_empty()
1708 
            || self
1709 
                .config
1710 
                .sparse_paths
1711 
                .iter()
1712 
                .any(|p| path == p || path.starts_with(&format!("{p}/")))
7421309f Andy McCall ↶ 1713 
    }
1714
98286406 Andy McCall ↶ 1715 
    fn walk_working_copy(&self) -> Result<BTreeMap<String, FsEntry>> {
696bbf29 Andy McCall ↶ 1716 
        self.walk_working_copy_with(None, &BTreeMap::new(), None)
4dee277c Andy McCall ↶ 1717 
    }
1718
b7f102fc Andy McCall ↶ 1719 
    /// Walk the working copy; with a manifest, a file whose size and mtime
cc7a0056 Andy McCall ↶ 1720 
    /// match it is not read: an opaque file yields its known blob, and a
1721 
    /// text file whose recorded graph key equals its base graph's key in
1722 
    /// `keys` (path → key) is marked unchanged; other text files are read
1723 
    /// but keep their known hash. Files modified within two seconds of the
1724 
    /// manifest are never trusted (a racy stat).
1725 
    ///
1726 
    /// `.oxyattributes` files are always read, since their content is
1727 
    /// needed; and when any of them is not known unchanged (added, removed,
1728 
    /// edited, or changed in the base), no text file is taken as unchanged,
1729 
    /// because attributes decide how a file compares with the base.
9ac1fbe2 Andy McCall ↶ 1730 
    ///
1731 
    /// `.oxyignore` applies to untracked paths only: every path in `tracked`
1732 
    /// (the base state's files) that exists on disk is included even when
1733 
    /// an ignore rule matches it, as in Git.
75167672 Andy McCall 1734 
    fn walk_working_copy_with(
1735 
        &self,
1736 
        cache: Option<&prompt::Manifest>,
1737 
        tracked: &BTreeMap<String, (NodeKind, Option<ObjectRef>)>,
992467d3 Andy McCall ↶ 1738 
        keys: Option<&KeyOf>,
75167672 Andy McCall 1739 
    ) -> Result<BTreeMap<String, FsEntry>> {
36e51aa8 Andy McCall ↶ 1740 
        let mut out: BTreeMap<String, FsEntry>;
76740263 Andy McCall ↶ 1741 
        let sparse = &self.config.sparse_paths;
8b276e01 Andy McCall ↶ 1742 
        let in_scope = |p: &str| {
1743 
            sparse.is_empty()
1744 
                || sparse
1745 
                    .iter()
1746 
                    .any(|s| p == s || p.starts_with(&format!("{s}/")))
1747 
        };
76740263 Andy McCall ↶ 1748 
        let walk_phase = crate::timing::phase("walk:   of which listing");
1749 
        let walk::Listing {
1750 
            files,
1751 
            symlinks,
1752 
            nested,
1ec3f008 Andy McCall ↶ 1753 
            dirs,
76740263 Andy McCall ↶ 1754 
        } = match self.listing_cache.take() {
1755 
            Some(l) => l,
d4dbab97 Andy McCall ↶ 1756 
            // Ignore rules apply to untracked paths only, so the base's
1757 
            // files are not matched against them (an added file is: a rule
1758 
            // written after it was snapshotted leaves it out again).
1759 
            None => walk::list_knowing(&self.root, &in_scope, &|p: &str| tracked.contains_key(p))?,
76740263 Andy McCall ↶ 1760 
        };
cc7a0056 Andy McCall ↶ 1761 
        let keys = keys.filter(|k| {
1762 
            let on_disk: HashSet<&str> = files.iter().map(|(r, _, _)| r.as_str()).collect();
1763 
            let attrs_unchanged =
1764 
                files
1765 
                    .iter()
1766 
                    .filter(|(r, _, _)| is_attributes(r))
1767 
                    .all(|(rel, _, meta)| {
1768 
                        stat_cached(rel, meta, cache)
992467d3 Andy McCall ↶ 1769 
                            .is_some_and(|e| e.graph.is_some() && e.graph == k(rel))
cc7a0056 Andy McCall ↶ 1770 
                    });
1771 
            let none_removed = tracked
1772 
                .keys()
1773 
                .filter(|r| is_attributes(r))
1774 
                .all(|r| on_disk.contains(r.as_str()));
1775 
            attrs_unchanged && none_removed
1776 
        });
1ec3f008 Andy McCall ↶ 1777 
        *self.walked_dirs.borrow_mut() =
1778 
            Some((dirs, nested.iter().map(|d| self.rel_path(d)).collect()));
76740263 Andy McCall ↶ 1779 
        drop(walk_phase);
1780 
        let _read_phase = crate::timing::phase("walk:   of which stat and compare");
e7ddd552 Andy McCall ↶ 1781 
        // Reading files is most of a walk's cost; do it in parallel.
1782 
        {
1783 
            use rayon::prelude::*;
36e51aa8 Andy McCall ↶ 1784 
            let mut read: Vec<(String, FsEntry)> = files
e7ddd552 Andy McCall ↶ 1785 
                .par_iter()
1786 
                .map(|(rel, path, meta)| {
992467d3 Andy McCall ↶ 1787 
                    // A key is worked out only for a file the manifest
1788 
                    // knows with the same stat and a key of its own.
cc7a0056 Andy McCall ↶ 1789 
                    let key = keys
1790 
                        .filter(|_| !is_attributes(rel))
992467d3 Andy McCall ↶ 1791 
                        .filter(|_| {
1792 
                            stat_cached(rel, meta, cache).is_some_and(|e| e.graph.is_some())
1793 
                        })
1794 
                        .and_then(|k| k(rel));
1966f5db Andy McCall ↶ 1795 
                    Ok((rel.clone(), read_file_entry(rel, path, meta, cache, key)?))
e7ddd552 Andy McCall ↶ 1796 
                })
1797 
                .collect::<Result<Vec<_>>>()?;
36e51aa8 Andy McCall ↶ 1798 
            for (rel, e) in read.iter_mut() {
e7ddd552 Andy McCall ↶ 1799 
                // A cached blob the store lost (gc, a partial clone): read the file after all.
1800 
                if e.content.is_empty() && e.blob.is_some_and(|b| !self.store.has(&b)) {
36e51aa8 Andy McCall ↶ 1801 
                    e.content = std::sync::Arc::new(fs::read(self.root.join(rel.as_str()))?);
e7ddd552 Andy McCall ↶ 1802 
                    e.blob = None;
1803 
                }
36e51aa8 Andy McCall ↶ 1804 
            }
1805 
            // The listing is in path order, so the map is built without
1806 
            // searching.
1807 
            out = read.into_iter().collect();
1808 
            for (rel, target) in symlinks {
1809 
                out.insert(rel, FsEntry::new(NodeKind::Symlink, false, target));
98286406 Andy McCall ↶ 1810 
            }
1811 
        }
76740263 Andy McCall ↶ 1812 
        drop(_read_phase);
1813 
        let _rest = crate::timing::phase("walk:   of which subrepos and hidden");
4dee277c Andy McCall ↶ 1814 
        // Nested repositories (spec §7.8): registered in `.oxymodules` or
1815 
        // already tracked as subrepo nodes, they contribute their pin;
1816 
        // unregistered ones are left alone.
1817 
        let modules = self.modules_on_disk();
1818 
        let removed = self.removed_subrepos();
1819 
        for rel in &removed {
1820 
            if tracked.get(rel).map(|(k, _)| *k) != Some(NodeKind::Subrepo) {
1821 
                self.clear_removed_subrepo(rel);
1822 
            }
1823 
        }
d4dbab97 Andy McCall ↶ 1824 
        let mut untracked_nested = Vec::new();
76740263 Andy McCall ↶ 1825 
        for dir in nested {
4dee277c Andy McCall ↶ 1826 
            let rel = self.rel_path(&dir);
1827 
            if !self.in_scope(&rel) || removed.contains(&rel) {
1828 
                continue;
1829 
            }
1830 
            let registered = modules.entries.get(&rel);
75167672 Andy McCall 1831 
            if registered.is_none() && tracked.get(&rel).map(|(k, _)| *k) != Some(NodeKind::Subrepo)
1832 
            {
d4dbab97 Andy McCall ↶ 1833 
                untracked_nested.push(rel);
4dee277c Andy McCall ↶ 1834 
                continue;
1835 
            }
1836 
            let pin = subrepos::read_pin(&dir, registered.and_then(|m| m.bookmark.as_deref()))?;
1837 
            out.insert(rel, FsEntry::new(NodeKind::Subrepo, false, pin.encode()));
1838 
        }
d4dbab97 Andy McCall ↶ 1839 
        untracked_nested.sort();
1840 
        *self.untracked_nested.borrow_mut() = untracked_nested;
9ac1fbe2 Andy McCall ↶ 1841 
        // Tracked paths the ignore rules hid.
f57bf670 Andy McCall ↶ 1842 
        for (rel, (kind, blob)) in missing_from(tracked, &out) {
1843 
            if !self.in_scope(rel) {
9ac1fbe2 Andy McCall ↶ 1844 
                continue;
1845 
            }
1846 
            let path = self.root.join(rel);
4dee277c Andy McCall ↶ 1847 
            if *kind == NodeKind::Subrepo {
1848 
                if removed.contains(rel) {
1849 
                    continue;
1850 
                }
75167672 Andy McCall 1851 
                let present =
1852 
                    path.is_dir() && (path.join(OXY_DIR).exists() || path.join(".git").exists());
4dee277c Andy McCall ↶ 1853 
                if present {
75167672 Andy McCall 1854 
                    let pin = subrepos::read_pin(
1855 
                        &path,
1856 
                        modules.entries.get(rel).and_then(|m| m.bookmark.as_deref()),
1857 
                    )?;
1858 
                    out.insert(
1859 
                        rel.clone(),
1860 
                        FsEntry::new(NodeKind::Subrepo, false, pin.encode()),
1861 
                    );
4dee277c Andy McCall ↶ 1862 
                } else if let Some(b) = blob {
1863 
                    // Not cloned (or emptied): the pin stays as recorded, as a
1864 
                    // Git gitlink does when its directory is absent.
75167672 Andy McCall 1865 
                    out.insert(
1866 
                        rel.clone(),
1867 
                        FsEntry {
1868 
                            blob: Some(*b),
cc7a0056 Andy McCall ↶ 1869 
                            ..FsEntry::new(NodeKind::Subrepo, false, vec![])
75167672 Andy McCall 1870 
                        },
1871 
                    );
4dee277c Andy McCall ↶ 1872 
                }
1873 
                continue;
1874 
            }
75167672 Andy McCall 1875 
            let Ok(meta) = fs::symlink_metadata(&path) else {
1876 
                continue;
1877 
            };
9ac1fbe2 Andy McCall ↶ 1878 
            let ft = meta.file_type();
1879 
            if ft.is_symlink() {
1880 
                let target = fs::read_link(&path)?;
75167672 Andy McCall 1881 
                out.insert(
1882 
                    rel.clone(),
1883 
                    FsEntry::new(
1884 
                        NodeKind::Symlink,
1885 
                        false,
1886 
                        target.to_string_lossy().as_bytes().to_vec(),
1887 
                    ),
1888 
                );
9ac1fbe2 Andy McCall ↶ 1889 
            } else if ft.is_file() {
e96a430c Andy McCall ↶ 1890 
                let e = self.file_entry(rel, &path, &walk::Stat::of(&meta), cache)?;
9ac1fbe2 Andy McCall ↶ 1891 
                out.insert(rel.clone(), e);
1892 
            }
1893 
        }
98286406 Andy McCall ↶ 1894 
        Ok(out)
1895 
    }
9ac1fbe2 Andy McCall ↶ 1896
1897 
    /// The entry for one regular file, read or served from the stat cache.
75167672 Andy McCall 1898 
    fn file_entry(
1899 
        &self,
1900 
        rel: &str,
1901 
        path: &Path,
e96a430c Andy McCall ↶ 1902 
        meta: &walk::Stat,
75167672 Andy McCall 1903 
        cache: Option<&prompt::Manifest>,
1904 
    ) -> Result<FsEntry> {
696bbf29 Andy McCall ↶ 1905 
        let mut e = read_file_entry(rel, path, meta, cache, None)?;
e7ddd552 Andy McCall ↶ 1906 
        if e.content.is_empty() && e.blob.is_some_and(|b| !self.store.has(&b)) {
40a40bb7 Andy McCall ↶ 1907 
            e.content = std::sync::Arc::new(fs::read(path)?);
e7ddd552 Andy McCall ↶ 1908 
            e.blob = None;
1909 
        }
1910 
        Ok(e)
9ac1fbe2 Andy McCall ↶ 1911 
    }
1912
4dee277c Andy McCall ↶ 1913 
    /// The non-directory paths of a snapshot with their kinds, for the
1914 
    /// walker's `tracked` set.
75167672 Andy McCall 1915 
    pub(crate) fn tracked_kinds(
1916 
        snapshot: &Snapshot,
1917 
    ) -> BTreeMap<String, (NodeKind, Option<ObjectRef>)> {
4dee277c Andy McCall ↶ 1918 
        snapshot
1919 
            .files
1920 
            .iter()
1921 
            .filter(|(_, f)| f.kind != NodeKind::Dir)
1922 
            .map(|(p, f)| {
1923 
                let blob = match &f.content {
1924 
                    FileContent::Opaque { blobs } => blobs.first().copied(),
1925 
                    _ => None,
1926 
                };
1927 
                (p.clone(), (f.kind, blob))
1928 
            })
1929 
            .collect()
1930 
    }
1931
1932 
    /// A path under the root as a `/`-separated relative string.
1933 
    pub(crate) fn rel_path(&self, path: &Path) -> String {
75167672 Andy McCall 1934 
        path.strip_prefix(&self.root)
1935 
            .unwrap_or(path)
1936 
            .components()
1937 
            .map(|c| c.as_os_str().to_string_lossy().to_string())
1938 
            .collect::<Vec<_>>()
1939 
            .join("/")
9ac1fbe2 Andy McCall ↶ 1940 
    }
98286406 Andy McCall ↶ 1941 
}
1942
1943 
impl Repo {
1944 
    /// Changes reachable from the bookmark and the working copy, newest first.
1945 
    pub fn log(&self) -> Result<Vec<LogEntry>> {
e02bf0cd Andy McCall ↶ 1946 
        let heads = self.bookmark_heads()?;
1947 
        if self.bare {
1948 
            return self.log_state(&heads);
1949 
        }
79b5a90f Andy McCall ↶ 1950 
        let (wc_hash, wc) = self.working_copy()?;
e02bf0cd Andy McCall ↶ 1951 
        let mut all = heads.clone();
1952 
        all.push(wc_hash);
1953 
        let head_set: BTreeSet<ChangeHash> = heads.into_iter().collect();
252e8cb2 Andy McCall ↶ 1954 
        let mut entries: Vec<LogEntry> = Vec::new();
4d6c9c1b Andy McCall ↶ 1955 
        // A shallow clone knows changes beyond its boundary from a
1956 
        // checkpoint's metadata only: its log comes from the state.
1957 
        let metas = if self.is_shallow() {
1958 
            self.load_state_history(&all)?
1959 
                .changes()
1960 
                .map(|(h, m)| (*h, m.clone()))
1961 
                .collect()
1962 
        } else {
79b5a90f Andy McCall ↶ 1963 
            // The working copy's change is in hand, and is never in the
1964 
            // state index: its metadata is not looked up again.
1965 
            let mut below: Vec<ChangeHash> = head_set.iter().copied().collect();
1966 
            below.extend(wc.deps.iter().copied());
1967 
            let _t = crate::timing::phase("log: metadata");
1968 
            let mut metas = self.metas_of(&below)?;
1969 
            if !metas.iter().any(|(h, _)| *h == wc_hash) {
1970 
                metas.push((wc_hash, crate::state::ChangeMeta::of(&wc)));
1971 
            }
1972 
            metas
4d6c9c1b Andy McCall ↶ 1973 
        };
1974 
        for (h, meta) in metas {
75167672 Andy McCall 1975 
            entries.push(LogEntry {
4d6c9c1b Andy McCall ↶ 1976 
                hash: h,
1977 
                change: meta,
1978 
                is_working_copy: h == wc_hash,
1979 
                is_head: head_set.contains(&h),
75167672 Andy McCall 1980 
            });
252e8cb2 Andy McCall ↶ 1981 
        }
79b5a90f Andy McCall ↶ 1982 
        let _sorting = crate::timing::phase("log: sort");
1983 
        // The hash makes the order total, so an unstable sort gives the same.
1984 
        entries.sort_unstable_by(|a, b| {
75167672 Andy McCall 1985 
            b.is_working_copy
1986 
                .cmp(&a.is_working_copy)
1987 
                .then(b.change.hlc.cmp(&a.change.hlc))
1988 
                .then(b.hash.cmp(&a.hash))
1989 
        });
98286406 Andy McCall ↶ 1990 
        Ok(entries)
1991 
    }
1992
119bbf50 Andy McCall ↶ 1993 
    /// A window of `log_state`: entries `offset..offset+limit` and the total
1994 
    /// count, for paginated views.
9d333fc9 Andy McCall ↶ 1995 
    pub fn log_window(
1996 
        &self,
1997 
        heads: &[ChangeHash],
1998 
        offset: usize,
1999 
        limit: usize,
2000 
    ) -> Result<(Vec<LogEntry>, usize)> {
119bbf50 Andy McCall ↶ 2001 
        let all = self.log_state(heads)?;
2002 
        let total = all.len();
2003 
        Ok((all.into_iter().skip(offset).take(limit).collect(), total))
2004 
    }
2005
e02bf0cd Andy McCall ↶ 2006 
    /// The log of the state with these heads, newest first, with no working
2007 
    /// copy involved (for hosted repositories and any explicit state).
2008 
    pub fn log_state(&self, heads: &[ChangeHash]) -> Result<Vec<LogEntry>> {
4d6c9c1b Andy McCall ↶ 2009 
        // The log needs change metadata only: neither graphs nor the tree.
2010 
        if self.is_shallow() {
2011 
            return Ok(self.log_of(&self.load_state_history(heads)?));
2012 
        }
2013 
        let metas = self.metas_of(heads)?;
2014 
        let depended: BTreeSet<ChangeHash> = metas
2015 
            .iter()
2016 
            .flat_map(|(_, m)| m.deps.iter().copied())
2017 
            .collect();
2018 
        let mut entries: Vec<LogEntry> = metas
2019 
            .into_iter()
2020 
            .map(|(h, meta)| LogEntry {
2021 
                hash: h,
2022 
                is_head: !depended.contains(&h),
2023 
                change: meta,
2024 
                is_working_copy: false,
2025 
            })
2026 
            .collect();
79b5a90f Andy McCall ↶ 2027 
        entries.sort_unstable_by(|a, b| b.change.hlc.cmp(&a.change.hlc).then(b.hash.cmp(&a.hash)));
4d6c9c1b Andy McCall ↶ 2028 
        Ok(entries)
2029 
    }
2030
2031 
    /// Every change of the state with these heads and its metadata, read
2032 
    /// without its operations ([`Store::get_change_meta`]). A change beyond
2033 
    /// a shallow clone's boundary is left out.
2034 
    pub fn metas_of(
2035 
        &self,
2036 
        heads: &[ChangeHash],
2037 
    ) -> Result<Vec<(ChangeHash, crate::state::ChangeMeta)>> {
2038 
        let boundary = self.boundary();
4cce912a Andy McCall ↶ 2039 
        // The state index holds the metadata of its heads' closure: a
2040 
        // change found there is not read from the store (a change object
2041 
        // can be as large as the tree it adds).
bf0247ee Andy McCall ↶ 2042 
        let read = self.read_state_index_changes();
2043 
        // Every one of `heads` in the index, and every head of the index
2044 
        // among them: their closure is the index's changes, all of them
2045 
        // (it is closed), and no walk is needed to find it.
2046 
        if let Some((index_heads, all)) = &read {
2047 
            let asked: crate::fasthash::FastSet<&ChangeHash> = heads.iter().collect();
2048 
            if heads.iter().all(|h| all.contains_key(h))
2049 
                && index_heads.iter().all(|h| asked.contains(h))
2050 
            {
2051 
                let (_, all) = read.expect("matched above");
2052 
                #[cfg(test)]
2053 
                {
2054 
                    // The walk the shortcut saves finds the same changes.
2055 
                    let mut seen: HashSet<ChangeHash> = HashSet::new();
2056 
                    let mut stack = heads.to_vec();
2057 
                    while let Some(h) = stack.pop() {
2058 
                        if seen.insert(h) {
2059 
                            stack.extend(self.store.get_change_meta(&h)?.deps);
2060 
                        }
2061 
                    }
2062 
                    assert_eq!(seen, all.keys().copied().collect::<HashSet<_>>());
2063 
                }
2064 
                return Ok(all
2065 
                    .into_iter()
2066 
                    .map(|(h, m)| (h, std::sync::Arc::unwrap_or_clone(m)))
2067 
                    .collect());
2068 
            }
2069 
        }
2070 
        let mut indexed = read.map(|(_, all)| all);
79b5a90f Andy McCall ↶ 2071 
        let _walk = crate::timing::phase("metadata: walk");
4cce912a Andy McCall ↶ 2072 
        let (mut stored, mut from_store) = (0usize, std::time::Duration::ZERO);
79b5a90f Andy McCall ↶ 2073 
        let mut seen: crate::fasthash::FastSet<ChangeHash> = Default::default();
4d6c9c1b Andy McCall ↶ 2074 
        let mut stack: Vec<ChangeHash> = heads.to_vec();
79b5a90f Andy McCall ↶ 2075 
        let mut out = Vec::with_capacity(indexed.as_ref().map_or(0, |m| m.len()));
4d6c9c1b Andy McCall ↶ 2076 
        while let Some(h) = stack.pop() {
2077 
            if !seen.insert(h) {
2078 
                continue;
2079 
            }
79b5a90f Andy McCall ↶ 2080 
            // Taken out of the index's map, each once: not copied.
2081 
            if let Some(meta) = indexed.as_mut().and_then(|m| m.remove(&h)) {
4cce912a Andy McCall ↶ 2082 
                #[cfg(test)]
2083 
                assert_eq!(
79b5a90f Andy McCall ↶ 2084 
                    *meta,
4cce912a Andy McCall ↶ 2085 
                    self.store.get_change_meta(&h)?,
2086 
                    "the state index holds a change's metadata"
2087 
                );
79b5a90f Andy McCall ↶ 2088 
                let meta = std::sync::Arc::unwrap_or_clone(meta);
4cce912a Andy McCall ↶ 2089 
                stack.extend(meta.deps.iter().copied());
79b5a90f Andy McCall ↶ 2090 
                out.push((h, meta));
4cce912a Andy McCall ↶ 2091 
                continue;
2092 
            }
2093 
            let started = std::time::Instant::now();
4d6c9c1b Andy McCall ↶ 2094 
            let meta = match self.store.get_change_meta(&h) {
2095 
                Ok(m) => m,
2096 
                Err(crate::store::StoreError::NotFound(_)) if boundary.contains(&h) => continue,
2097 
                Err(e) => return Err(e.into()),
2098 
            };
4cce912a Andy McCall ↶ 2099 
            stored += 1;
2100 
            from_store += started.elapsed();
4d6c9c1b Andy McCall ↶ 2101 
            stack.extend(meta.deps.iter().copied());
2102 
            out.push((h, meta));
4cce912a Andy McCall ↶ 2103 
        }
2104 
        if stored > 0 && crate::timing::enabled() {
2105 
            eprintln!("timing metadata: {stored} changes read whole, {from_store:.2?}");
4d6c9c1b Andy McCall ↶ 2106 
        }
2107 
        Ok(out)
bc489b6c Andy McCall ↶ 2108 
    }
2109
2110 
    /// The log of a state already loaded, newest first.
83d61e80 Andy McCall ↶ 2111 
    /// The history of lines `first..=last` (1-based) of `path` in the
2112 
    /// state `heads` make (spec §27): every change that inserted or
2113 
    /// deleted one of their atoms, or an atom deleted from between them,
2114 
    /// newest first. Exact, as atoms keep their identity across edits.
2115 
    pub fn line_history(
2116 
        &self,
2117 
        heads: &[ChangeHash],
2118 
        path: &str,
2119 
        first: usize,
2120 
        last: usize,
2121 
    ) -> Result<Vec<LogEntry>> {
2122 
        if first == 0 || last < first {
2123 
            return fail(format!(
2124 
                "lines {first},{last}: expected FIRST,LAST from 1 with FIRST ≤ LAST"
2125 
            ));
2126 
        }
2127 
        let mut state = self.load_state_history(heads)?;
2128 
        let Some(node) = state.tree().view().entries.get(path).map(|e| e.node) else {
2129 
            return fail(format!("{path} is not in that state"));
2130 
        };
2131 
        self.load_graphs_of(&mut state, &[node])?;
2132 
        let live = state.live_lines(&node);
2133 
        if last > live.len() {
2134 
            return fail(format!(
2135 
                "{path} has {} lines; asked for {first},{last}",
2136 
                live.len()
2137 
            ));
2138 
        }
2139 
        let Some(g) = state.file(&node) else {
2140 
            return fail(format!("{path} is not a text file"));
2141 
        };
2142 
        let atoms = g.between(&live[first - 1].0, &live[last - 1].0);
2143 
        let touched = g.touching(&atoms);
2144 
        Ok(self
2145 
            .log_of(&state)
2146 
            .into_iter()
2147 
            .filter(|e| touched.contains(&e.hash))
2148 
            .collect())
2149 
    }
2150
bc489b6c Andy McCall ↶ 2151 
    pub fn log_of(&self, state: &crate::state::State) -> Vec<LogEntry> {
e02bf0cd Andy McCall ↶ 2152 
        let head_set: BTreeSet<ChangeHash> = state.heads().into_iter().collect();
2153 
        let mut entries: Vec<LogEntry> = state
2154 
            .changes()
75167672 Andy McCall 2155 
            .map(|(h, meta)| LogEntry {
2156 
                hash: *h,
2157 
                change: meta.clone(),
2158 
                is_working_copy: false,
2159 
                is_head: head_set.contains(h),
2160 
            })
e02bf0cd Andy McCall ↶ 2161 
            .collect();
79b5a90f Andy McCall ↶ 2162 
        entries.sort_unstable_by(|a, b| b.change.hlc.cmp(&a.change.hlc).then(b.hash.cmp(&a.hash)));
bc489b6c Andy McCall ↶ 2163 
        entries
e02bf0cd Andy McCall ↶ 2164 
    }
2165
14edbaaf Andy McCall ↶ 2166 
    /// Resolve a spec to a set of heads: a bookmark's name, a remote's
2167 
    /// bookmark as `REMOTE/NAME`, a tag name, or a single change
2168 
    /// (`resolve_change`).
132b884b Andy McCall ↶ 2169 
    pub fn resolve_heads(&self, spec: &str) -> Result<Vec<ChangeHash>> {
7b9809e6 Andy McCall ↶ 2170 
        let view = self.view()?;
2171 
        if let Some(st) = view.bookmarks.get(spec) {
2172 
            return Ok(self.store.get_state(st)?.heads);
132b884b Andy McCall ↶ 2173 
        }
14edbaaf Andy McCall ↶ 2174 
        if let Some((_, _, st)) = remote_bookmark(&view, spec) {
2175 
            return Ok(self.store.get_state(&st)?.heads);
2176 
        }
132b884b Andy McCall ↶ 2177 
        if let Some(heads) = self.tag_heads(spec)? {
2178 
            return Ok(heads);
2179 
        }
2180 
        Ok(vec![self.resolve_change(spec)?])
2181 
    }
2182
6a81dbab Andy McCall ↶ 2183 
    /// Resolve a change by ChangeId prefix, `#hash` prefix, `@` (working
14edbaaf Andy McCall ↶ 2184 
    /// copy), bookmark, `REMOTE/NAME` or tag (the one head of its state).
98286406 Andy McCall ↶ 2185 
    pub fn resolve_change(&self, spec: &str) -> Result<ChangeHash> {
2186 
        if spec == "@" {
2187 
            return Ok(self.working_copy()?.0);
2188 
        }
14edbaaf Andy McCall ↶ 2189 
        let view = self.view()?;
2190 
        let named = match view.bookmarks.get(spec) {
6a81dbab Andy McCall ↶ 2191 
            Some(st) => Some(("bookmark", self.store.get_state(st)?.heads)),
14edbaaf Andy McCall ↶ 2192 
            None => match remote_bookmark(&view, spec) {
2193 
                Some((_, _, st)) => Some(("bookmark", self.store.get_state(&st)?.heads)),
2194 
                None => self.tag_heads(spec)?.map(|h| ("tag", h)),
2195 
            },
6a81dbab Andy McCall ↶ 2196 
        };
2197 
        if let Some((kind, heads)) = named {
7b9809e6 Andy McCall ↶ 2198 
            return match heads.len() {
2199 
                1 => Ok(heads[0]),
6a81dbab Andy McCall ↶ 2200 
                n => fail(format!("{kind} {spec} has {n} heads; name a change, or use it where a state is expected")),
7b9809e6 Andy McCall ↶ 2201 
            };
2202 
        }
98286406 Andy McCall ↶ 2203 
        if spec.is_empty() || spec == "#" {
2204 
            return fail("empty change specifier");
2205 
        }
2206 
        let entries = self.log()?;
2207 
        let matches: Vec<&LogEntry> = if let Some(hp) = spec.strip_prefix('#') {
75167672 Andy McCall 2208 
            entries
2209 
                .iter()
2210 
                .filter(|e| e.hash.starts_with_hex(hp))
2211 
                .collect()
98286406 Andy McCall ↶ 2212 
        } else {
75167672 Andy McCall 2213 
            entries
2214 
                .iter()
2215 
                .filter(|e| e.change.change_id.starts_with_hex(spec))
2216 
                .collect()
98286406 Andy McCall ↶ 2217 
        };
2218 
        match matches.len() {
2219 
            1 => Ok(matches[0].hash),
b35df36b Andy McCall ↶ 2220 
            0 => match self.resolve_in_remotes(spec)? {
2221 
                Some(h) => Ok(h),
2222 
                None => fail(format!("no change matches {spec:?}")),
2223 
            },
98286406 Andy McCall ↶ 2224 
            _ => {
2225 
                let ids: BTreeSet<ChangeId> = matches.iter().map(|e| e.change.change_id).collect();
2226 
                if ids.len() == 1 {
75167672 Andy McCall 2227 
                    fail(format!(
2228 
                        "change {spec} is divergent: {} variants",
2229 
                        matches.len()
2230 
                    ))
98286406 Andy McCall ↶ 2231 
                } else {
2232 
                    fail(format!("{spec:?} is ambiguous"))
b35df36b Andy McCall ↶ 2233 
                }
2234 
            }
2235 
        }
2236 
    }
2237
2238 
    /// A change no local bookmark holds, found by id or `#hash` prefix in
2239 
    /// the remote bookmarks' states: fetched but not yet pulled (oxygen#56).
2240 
    /// `None` when none matches, or more than one.
2241 
    fn resolve_in_remotes(&self, spec: &str) -> Result<Option<ChangeHash>> {
2242 
        let view = self.view()?;
2243 
        let mut found: BTreeSet<ChangeHash> = BTreeSet::new();
2244 
        for bookmarks in view.remotes.values() {
2245 
            for st in bookmarks.values() {
2246 
                let heads = self.store.get_state(st)?.heads;
2247 
                for e in self.log_state(&heads)? {
2248 
                    let hit = match spec.strip_prefix('#') {
2249 
                        Some(hp) => e.hash.starts_with_hex(hp),
2250 
                        None => e.change.change_id.starts_with_hex(spec),
2251 
                    };
2252 
                    if hit {
2253 
                        found.insert(e.hash);
2254 
                    }
2255 
                }
2256 
            }
2257 
        }
2258 
        Ok((found.len() == 1).then(|| *found.first().unwrap()))
2259 
    }
2260
2261 
    /// Where a change is when no local bookmark or the working copy holds
2262 
    /// it: the first remote bookmark that does (`origin/main`). `None` when
2263 
    /// it is held locally, or nowhere.
2264 
    pub fn remote_only_location(&self, hash: &ChangeHash) -> Result<Option<String>> {
2265 
        if self.log()?.iter().any(|e| e.hash == *hash) {
2266 
            return Ok(None);
2267 
        }
2268 
        let view = self.view()?;
2269 
        for (remote, bookmarks) in &view.remotes {
2270 
            for (name, st) in bookmarks {
2271 
                let heads = self.store.get_state(st)?.heads;
2272 
                if self.log_state(&heads)?.iter().any(|e| e.hash == *hash) {
2273 
                    return Ok(Some(format!("{remote}/{name}")));
98286406 Andy McCall ↶ 2274 
                }
2275 
            }
2276 
        }
b35df36b Andy McCall ↶ 2277 
        Ok(None)
98286406 Andy McCall ↶ 2278 
    }
2279
2280 
    /// Human-readable description of what a change does, with line content.
2281 
    pub fn describe_change(&self, hash: &ChangeHash) -> Result<String> {
2282 
        let change = self.store.get_change(hash)?;
8ab265dc Andy McCall ↶ 2283 
        let (base, base_snapshot) = self.change_base(&change)?;
98286406 Andy McCall ↶ 2284 
        Ok(render_change(&change, &base, &base_snapshot))
8ab265dc Andy McCall ↶ 2285 
    }
2286
2287 
    /// The base state of `change` (its dependencies) and its rendering, as
2288 
    /// much as showing the change needs: the whole tree, and the graphs and
2289 
    /// contents of the files the change edits. Other text files are left
2290 
    /// unrendered ([`Snapshot::unrendered`]).
2291 
    pub fn change_base(&self, change: &Change) -> Result<(State, Snapshot)> {
77445673 Andy McCall ↶ 2292 
        let base = self.load_state_history(&change.deps)?;
2293 
        self.render_base_for(change, base)
2294 
    }
2295
dd6147dc Andy McCall ↶ 2296 
    /// The base for showing `change` alone: the graphs of the files it
2297 
    /// edits built, and the base's tree with no file rendered, which is
2298 
    /// all [`render_change`](snapshot::render_change) and
2299 
    /// [`diff_files_of`](snapshot::diff_files_of) read.
2300 
    pub(crate) fn tree_base_for(
2301 
        &self,
2302 
        change: &Change,
2303 
        mut base: State,
2304 
    ) -> Result<(State, Snapshot)> {
2305 
        let _t = crate::timing::phase("diff: the base's tree");
2306 
        let edited: std::collections::HashSet<NodeId> =
2307 
            change.edits.iter().map(|e| e.node()).collect();
2308 
        let nodes: Vec<NodeId> = edited.into_iter().collect();
2309 
        self.load_graphs_of(&mut base, &nodes)?;
2310 
        let snapshot = Snapshot {
2311 
            tree: base.tree().view(),
2312 
            files: BTreeMap::new(),
2313 
            unrendered: BTreeSet::new(),
2314 
        };
2315 
        Ok((base, snapshot))
2316 
    }
2317
77445673 Andy McCall ↶ 2318 
    /// [`Repo::change_base`] with the base already loaded (the working
2319 
    /// copy's comes from the state index).
2320 
    pub(crate) fn render_base_for(
2321 
        &self,
2322 
        change: &Change,
2323 
        mut base: State,
2324 
    ) -> Result<(State, Snapshot)> {
10b4fd43 Andy McCall ↶ 2325 
        let _t = crate::timing::phase("diff: render the base");
8ab265dc Andy McCall ↶ 2326 
        let edited: std::collections::HashSet<NodeId> =
2327 
            change.edits.iter().map(|e| e.node()).collect();
2328 
        let nodes: Vec<NodeId> = edited.iter().copied().collect();
2329 
        self.load_graphs_of(&mut base, &nodes)?;
10b4fd43 Andy McCall ↶ 2330 
        let snapshot = {
2331 
            let _t = crate::timing::phase("diff:   of which materialize");
2332 
            base.materialize_nodes(&edited)
2333 
        };
8ab265dc Andy McCall ↶ 2334 
        Ok((base, snapshot))
98286406 Andy McCall ↶ 2335 
    }
2336 
}
2337
2338 
fn remove_any(path: &Path) -> std::io::Result<()> {
2339 
    let m = fs::symlink_metadata(path)?;
75167672 Andy McCall 2340 
    if m.is_dir() {
2341 
        fs::remove_dir_all(path)
2342 
    } else {
2343 
        fs::remove_file(path)
2344 
    }
98286406 Andy McCall ↶ 2345 
}
2346
cdb5beba Andy McCall ↶ 2347 
fn set_readonly(path: &Path, readonly: bool) -> std::io::Result<()> {
2348 
    let mut perms = fs::metadata(path)?.permissions();
2349 
    perms.set_readonly(readonly);
2350 
    fs::set_permissions(path, perms)
2351 
}
2352
98286406 Andy McCall ↶ 2353 
#[cfg(unix)]
2354 
fn set_executable(path: &Path, exec: bool) -> std::io::Result<()> {
2355 
    use std::os::unix::fs::PermissionsExt;
2356 
    let mut perms = fs::metadata(path)?.permissions();
2357 
    let mode = perms.mode();
2358 
    let want = if exec { mode | 0o111 } else { mode & !0o111 };
2359 
    if want != mode {
2360 
        perms.set_mode(want);
2361 
        fs::set_permissions(path, perms)?;
2362 
    }
2363 
    Ok(())
2364 
}
2365 
#[cfg(not(unix))]
2366 
fn set_executable(_path: &Path, _exec: bool) -> std::io::Result<()> {
2367 
    Ok(())
2368 
}
2369
2370 
#[cfg(unix)]
e96a430c Andy McCall ↶ 2371 
pub(crate) fn is_executable(m: &fs::Metadata) -> bool {
98286406 Andy McCall ↶ 2372 
    use std::os::unix::fs::PermissionsExt;
2373 
    m.permissions().mode() & 0o111 != 0
2374 
}
2375 
#[cfg(not(unix))]
e96a430c Andy McCall ↶ 2376 
pub(crate) fn is_executable(_m: &fs::Metadata) -> bool {
98286406 Andy McCall ↶ 2377 
    false
2378 
}
2379
2380 
fn op_ref(k: OpKey) -> OpRef {
75167672 Andy McCall 2381 
    OpRef {
2382 
        change: k.0,
2383 
        index: k.1,
2384 
    }
98286406 Andy McCall ↶ 2385 
}
2386
2387 
fn split_path(path: &str) -> (&str, &str) {
2388 
    match path.rfind('/') {
2389 
        Some(i) => (&path[..i], &path[i + 1..]),
2390 
        None => ("", path),
2391 
    }
cc7a0056 Andy McCall ↶ 2392 
}
2393
2394 
/// Whether `rel` is an attributes file (spec §7.6).
2395 
fn is_attributes(rel: &str) -> bool {
2396 
    rel == ".oxyattributes" || rel.ends_with("/.oxyattributes")
98286406 Andy McCall ↶ 2397 
}
e7ddd552 Andy McCall ↶ 2398
cc7a0056 Andy McCall ↶ 2399 
/// The manifest entry for `rel` when its size and mtime match it outside
2400 
/// the racy window.
2401 
fn stat_cached<'m>(
2402 
    rel: &str,
e96a430c Andy McCall ↶ 2403 
    meta: &walk::Stat,
cc7a0056 Andy McCall ↶ 2404 
    cache: Option<&'m prompt::Manifest>,
2405 
) -> Option<&'m prompt::ManifestEntry> {
2406 
    let m = cache?;
2407 
    let e = m.entries.get(rel)?;
e96a430c Andy McCall ↶ 2408 
    let racy = meta.mtime_ns + 2_000_000_000 >= m.written_ns;
2409 
    (e.size == meta.size && e.mtime_ns == meta.mtime_ns && !racy).then_some(e)
cc7a0056 Andy McCall ↶ 2410 
}
2411
2412 
/// Read one working-copy file, or serve it from the stat cache (spec
2413 
/// §12.3). With a manifest whose stat for it is unchanged (and not racy),
2414 
/// an opaque file is not read at all (its stored blob stands in), and a
2415 
/// text file the manifest recorded equal to a base graph whose key is
2416 
/// `key`, the key the base has now, is not read either: equal keys mean
2417 
/// equal renderings, so it still equals the base. Touches no repository
2418 
/// state, so the walk runs it in parallel; the caller checks that a cached
2419 
/// blob is still in the store.
e7ddd552 Andy McCall ↶ 2420 
fn read_file_entry(
2421 
    rel: &str,
2422 
    path: &Path,
e96a430c Andy McCall ↶ 2423 
    meta: &walk::Stat,
e7ddd552 Andy McCall ↶ 2424 
    cache: Option<&prompt::Manifest>,
1966f5db Andy McCall ↶ 2425 
    key: Option<[u8; 32]>,
e7ddd552 Andy McCall ↶ 2426 
) -> Result<FsEntry> {
e96a430c Andy McCall ↶ 2427 
    let exec = meta.exec;
2428 
    let stat = Some((meta.size, meta.mtime_ns));
8b276e01 Andy McCall ↶ 2429 
    let mut e = match stat_cached(rel, meta, cache) {
e7ddd552 Andy McCall ↶ 2430 
        Some(e) if e.blob.is_some() => FsEntry {
2431 
            blob: e.blob,
1966f5db Andy McCall ↶ 2432 
            content_hash: Some(e.hash),
cc7a0056 Andy McCall ↶ 2433 
            ..FsEntry::new(NodeKind::File, exec, vec![])
2434 
        },
1966f5db Andy McCall ↶ 2435 
        Some(e) if key.is_some() && e.graph == key => FsEntry {
2436 
            content_hash: Some(e.hash),
cc7a0056 Andy McCall ↶ 2437 
            unchanged: true,
1966f5db Andy McCall ↶ 2438 
            graph_key: e.graph,
cc7a0056 Andy McCall ↶ 2439 
            ..FsEntry::new(NodeKind::File, exec, vec![])
2440 
        },
2441 
        Some(e) => FsEntry {
1966f5db Andy McCall ↶ 2442 
            content_hash: Some(e.hash),
cc7a0056 Andy McCall ↶ 2443 
            ..FsEntry::new(NodeKind::File, exec, fs::read(path)?)
e7ddd552 Andy McCall ↶ 2444 
        },
2445 
        None => FsEntry::new(NodeKind::File, exec, fs::read(path)?),
8b276e01 Andy McCall ↶ 2446 
    };
2447 
    // The stat the manifest records: taken before the read, so a file that
2448 
    // changes meanwhile is read again next time.
2449 
    e.stat = stat;
2450 
    Ok(e)
e7ddd552 Andy McCall ↶ 2451 
}
f57bf670 Andy McCall ↶ 2452
2453 
/// Two maps side by side in key order: each key of either, with its value
2454 
/// in each. Linear, with no lookups, for maps of a whole tree.
2455 
pub(crate) fn join<'a, A, B>(
2456 
    a: &'a BTreeMap<String, A>,
2457 
    b: &'a BTreeMap<String, B>,
2458 
) -> impl Iterator<Item = (&'a String, Option<&'a A>, Option<&'a B>)> {
2459 
    let (mut a, mut b) = (a.iter().peekable(), b.iter().peekable());
2460 
    std::iter::from_fn(move || match (a.peek(), b.peek()) {
2461 
        (None, None) => None,
2462 
        (Some(_), None) => a.next().map(|(k, v)| (k, Some(v), None)),
2463 
        (None, Some(_)) => b.next().map(|(k, v)| (k, None, Some(v))),
2464 
        (Some((ka, _)), Some((kb, _))) => match ka.cmp(kb) {
2465 
            std::cmp::Ordering::Less => a.next().map(|(k, v)| (k, Some(v), None)),
2466 
            std::cmp::Ordering::Greater => b.next().map(|(k, v)| (k, None, Some(v))),
2467 
            std::cmp::Ordering::Equal => {
2468 
                let (k, va) = a.next()?;
2469 
                let (_, vb) = b.next()?;
2470 
                Some((k, Some(va), Some(vb)))
2471 
            }
2472 
        },
2473 
    })
2474 
}
2475
2476 
/// The entries of `a` whose keys `b` lacks, in key order: linear, with no
2477 
/// lookups.
2478 
pub(crate) fn missing_from<'a, A, B>(
2479 
    a: &'a BTreeMap<String, A>,
2480 
    b: &BTreeMap<String, B>,
2481 
) -> Vec<(&'a String, &'a A)> {
2482 
    let mut keys = b.keys().peekable();
2483 
    a.iter()
2484 
        .filter(|(k, _)| {
2485 
            while keys.next_if(|kb| *kb < *k).is_some() {}
2486 
            keys.peek() != Some(k)
2487 
        })
2488 
        .collect()
2489 
}